NSF Driver and Export Fixes

[Gumball2415]

This pull request aims to fix various bugs regarding NSF driver export and execution.

This pull request improves upon and supercedes #156.

Changes in this PR:

• Major export refactoring
  • Use CCompiler pointer for writing ASM export title
  • Decouple NSFDRV header from driver data
  • Add NSF stub file for .asm export
  • Fix assembly period table export
  • Fix VRC7 period table assembly export
  • Implement "glob" multichip assembly export
  • Ignore invalid pattern data upon NSF export
    • Fixes NSF Export Issues in 0.5.0.1 #215
  • Specify what parts of pattern data are invalid
  • Differentiate between error and warning colors
  • Account for HOLD_INSTRUMENT in pattern checking
  • Specify overflow in driver bank allocation
  • Refactor compressed mode size calculation
  • Include N163 period table in .asm export
  • Declare alignment for first DPCM sample in assembly export
  • Remove hardcoded bank references in assembly song header
  • Implement bankswitched assembly config exports
  • Add bank 0 for nonbankswitching config
  • Refactor NSFe metadata generation
  • Fix NES ROM export data alignment
    • Fixes NROM iNES Export: Fix launcher kernel #278
  • Fix DPCM segment start address in nonbankswitching assembly
  • Refactor multifile export
• NSF driver fixes
  • Fix N163 instrument load clobbering Kxx state
    • Fixes NSF driver: N163 instrument load clobbering Kxx state #224.
    • Fixes NSF export sometimes skips some N163 channels #260.
    • Iterates on Update NSF Driver and NSF-related things #156
  • Fix S5B init clobbering Kxx state
    • Fixes 2A03+VRC6+MMC5 NSF export makes 2A03 pulse 1 play only pitch A-1 #256.
    • Fixes [S5B] 2A03's Pulse0 channel does not work properly when exporting an NSF with the S5B chip expansion. #246.
    • Fixes NSF Export Issues in 0.5.0.1 #215.
  • Write carrier frequency regardless of Kxx state
    • Fixes Issue with FDS modulation not being exported to NSF #196.
  • Fix phase reset behavior
  • Refactor phase reset code
  • Update effect bytecodes in row skip handling
    • Fixes Dxx does not behave properly in NSF export (under specific circumstances) #183.
  • Handle Gxx delay commands in both row and frame transitions
    • Fixes NSF export effects state corruption? #207.
  • Fix S5B instrument duty envelopes
    • Fixes S5B envelopes broken on NSF export #267.
  • Clamp pitch and hi-pitch envelope period results
  • Enable pitch sliding in linear pitch mode
  • Implement linear pitch period limit
  • Correct linear pitch calculation sequence
  • Fix length counter load not being written with pitch
    • Fixes Hardware sweep (Hxx + Ixx) and envelope (EEx) not working together in NSF export #297.
  • Prevent volume commands in DPCM channel to be written in NSF export
    • Fixes In NSF export, DPCM notes reset Vxx duty cycle of Pulse 1 #311.
• Internal fixes
  • Fix debug constexpr
• In-beta fixes:
  • Refactor Pitch/HiPitch bend logic
  • Apply N163 slide shift if linear pitch is disabled
    • Fixes (dev branch) N163's pitch slide speed is slower in NSF export compared to Famitracker playback #276.
  • Refactor period limit subroutine
    • Fixes (dev branch) N163's pitch slide speed is slower in NSF export compared to Famitracker playback #276.
  • Fix N163 portamento shift in linear period
    • Fixes (dev branch) N163's pitch slide speed is slower in NSF export compared to Famitracker playback #276.
  • Include .asm filename in NSF stub file
    • Fixes (Beta) NSF stub file doesn't include actual .asm file in .include directive #298.
  • Fix N163 Vxx indexing out of bounds
    • Fixes (beta) NSF driver N163 Vxx wave index can access out-of-bounds memory #299.

Leafuw, nyanpasu64, and Zeta, thank you for your help!

[Gumball2415]

TODO: fix linear pitch pitch sequences

fixed in 4ce047d

TODO: fix asm export DPCM start address issue with beep.dnm

fixed in fa05bd9

[nyanpasu64]

• is ChunkRenderText used for assembly export? i have not tested it, do you know it works for generating nsfs or games?
• // str.Format("; Bank %i\n", pChunk->GetBank()); generally when making a PR i usually delete code or clarify why it's not needed, instead of commenting out (which can confuse the next reader what the code is there for)
• i do not know enough of the code to provide meaningful feedback
• what is pChunk->StoreByte(m_iWaveCount);? does this allow the runtime driver to bounds-check an instrument? seems the driver was modified to take into account, at comment check if within actual wave count.
• PatternCompiler.cpp is giving me flashbacks to (invalid) notes popping up in 0cc 0.3.15.x when it corrupted its own memory
  • did you both turn Print into a template that calls _sntprintf_s internally (the more characters you put around printf the more powerful it is 😉), and only call it with CString::Format() single strings? in theory this would be dangerous if a % sign pops up in the string and the sprintf-like call thinks it's a format specifier and corrupts memory instead, but in practice it seems you only call the function with error messages populated with just integers.
• have not looked through the pattern editor code, seems you replaced (invalid) with question marks?

[Gumball2415]

responses in the same list order:

• CChunkRenderText handles writing the module data into a CA65 assembly representation (much like how CChunkRenderBinary handles writing module data into compiled PRG bytes for .bin)
• noted, will amend this.
• not much comment, other than this is also a good reason why documentation should be improved
• this stores the amount of waves an N163 instrument contains to help with bounds checking when indexing wave data. this was commented out in the source code of both the tracker and NSF driver as early as vanilla 0.4.0.
• i forgot why i changed it like that, maybe just a hack for formatted string printing?
  • edit: this implementation is directly copied from ExportDialog.cpp. all it does is replace LF line endings with CRLF for MFC.
  • this will be part of an eventual goal to substitute CString with std::string whenever possible (Replace CString with std::string #258)
• i changed it from printing (invalid) to ? to fit the text within each note, instrument, volume, and effect command.
  • more work needs to be done to create a more comprehensive and centralized pattern validator ([Feature] Pattern data validator #312)