WASM fixes

[uncomputable]

Fixes issues that came up while parsing the human encoding in WASM.

[apoelstra]

thread_local is more efficient than Once because it doesn't need to use synchronization every time that it's accessed.

[apoelstra]

If WASM can't handle thread_local, even though I wrote the original code specifically with WASM in mind (in which it should be a no-op because WASM has no threads), then I guess it's fine to hack around it by making our code less efficient and unsafe. But I'm not thrilled about it.

[apoelstra]

I can't test this until we update the vendored version of Simplicity for the recent CI fixes.

[apoelstra]

Actually, no, I was just building from the wrong system and had an outdated check-pr.nix. I think this is working now.

[apoelstra]

In 8355454:

This test fails on my system with

failures:

---- human_encoding::parse::tests::const_word stdout ----
thread 'human_encoding::parse::tests::const_word' panicked at 'byte index 2 is out of bounds of `0`', src/human_encoding/parse/ast.rs:365:47
stack backtrace:
   0: rust_begin_unwind
             at /rustc/eb26296b556cef10fb713a38f3d16b9886080f26/library/std/src/panicking.rs:593:5
   1: core::panicking::panic_fmt
             at /rustc/eb26296b556cef10fb713a38f3d16b9886080f26/library/core/src/panicking.rs:67:14
   2: core::str::slice_error_fail_rt
   3: core::str::slice_error_fail
             at /rustc/eb26296b556cef10fb713a38f3d16b9886080f26/library/core/src/str/mod.rs:87:9
   4: core::str::traits::<impl core::slice::index::SliceIndex<str> for core::ops::range::Range<usize>>::index
             at /rustc/eb26296b556cef10fb713a38f3d16b9886080f26/library/core/src/str/traits.rs:235:21
   5: core::str::traits::<impl core::ops::index::Index<I> for str>::index
             at /rustc/eb26296b556cef10fb713a38f3d16b9886080f26/library/core/src/str/traits.rs:61:9
   6: simplicity::human_encoding::parse::ast::Ast<J>::from_literal_lexeme
             at ./src/human_encoding/parse/ast.rs:365:47
   7: core::ops::function::Fn::call
             at /rustc/eb26296b556cef10fb713a38f3d16b9886080f26/library/core/src/ops/function.rs:79:5
   8: santiago::parser::tree::Tree<AST>::as_abstract_syntax_tree
             at /store/home/apoelstra/.cargo/registry/src/index.crates.io-6f17d22bba15001f/santiago-1.3.1/src/parser/tree.rs:91:46
   9: simplicity::human_encoding::parse::ast::parse_line_vector
             at ./src/human_encoding/parse/ast.rs:122:11
  10: simplicity::human_encoding::parse::parse
             at ./src/human_encoding/parse/mod.rs:180:17
  11: simplicity::human_encoding::parse::tests::assert_const
             at ./src/human_encoding/parse/mod.rs:642:15
  12: simplicity::human_encoding::parse::tests::const_word
             at ./src/human_encoding/parse/mod.rs:774:13
  13: simplicity::human_encoding::parse::tests::const_word::{{closure}}
             at ./src/human_encoding/parse/mod.rs:751:21
  14: core::ops::function::FnOnce::call_once
             at /rustc/eb26296b556cef10fb713a38f3d16b9886080f26/library/core/src/ops/function.rs:250:5
  15: core::ops::function::FnOnce::call_once
             at /rustc/eb26296b556cef10fb713a38f3d16b9886080f26/library/core/src/ops/function.rs:250:5
note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace.

I think you just need to reorder the commit that introduces tests with the one that fixes parsing of single-character hex literals.

[uncomputable]

Yes, I will reorder the commits. I will also try to fix the precomputed powers of two without using Once.

[uncomputable]

For now, I reordered the commits to make the tests work.

[apoelstra]

ACK 97e41f5

[uncomputable]

thread_local! works, so I added it back in. The borrow error no longer occurs.

We could share the same powers of two across threads using RwLock or OnceLock, but this requires MSRV 1.63.0 and MSRV 1.70.0, respectively, to work in const, so this is for later.

[apoelstra]

RwLock is extremely slow and should basically never be used unless your problem architecture really requires it. Not sure about OnceLock. It's likely to be better.

[apoelstra]

ACK c57d8de