Grammar based moves parsing

[everyonesdesign]

A better, more flexible way to parse moves defined by grammar instead of regular expressions

Using Nearley to generate and parse the grammar

TypeScript definitions are manually created

Potentially can unblock things like #74

Concerns

Performance and the bundle size.

Bundle size ✅

Bundle size (master): 116K
Bundle size (feature): 126K

Performance ❌

It seems that the performance with this parsing decreases significantly:

Master
  Execution time (1000 cycles, e): 1.375 ms
  Execution time (1000 cycles, e4): 1.7349999845027924 ms
  Execution time (1000 cycles, e4): 1.3700000047683716 ms
  Execution time (1000 cycles, d): 0.75 ms
  Execution time (1000 cycles, d5): 1.145000010728836 ms
  Execution time (1000 cycles, d5): 1.4149999916553497 ms
  Execution time (1000 cycles, d): 0.6149999797344208 ms
  Execution time (1000 cycles, d5): 1.7850000262260437 ms
  Execution time (1000 cycles, d5): 0.8899999856948853 ms

Feature
  Execution time (1000 cycles, e): 11.324999988079071 ms
  Execution time (1000 cycles, e4): 34.96999999880791 ms
  Execution time (1000 cycles, e4): 38.46999999880791 ms
  Execution time (1000 cycles, d): 5.875 ms
  Execution time (1000 cycles, d5): 25.589999973773956 ms
  Execution time (1000 cycles, d5): 30.680000007152557 ms
  Execution time (1000 cycles, d): 4.925000011920929 ms
  Execution time (1000 cycles, d5): 24.645000010728836 ms
  Execution time (1000 cycles, d5): 25.5 ms

[everyonesdesign]

Closing in favour of #82

Implementation based on Nearley seems to be difficult to maintain and very slow.