[Parser] Issues to resolve before switching to the new wat parser

[tlively]

These are the issues (in no particular order) that need to be resolved before we can turn on the new wat parser by default:

• The parser needs to support bottom and exn reftypes and heaptypes.
  • [Parser] Parse remaining heap and reference types #6218
• The parser needs to support start functions.
  • [Parser] Parse start declarations #6256
• The parser needs to support v128.const.
  • [Parser] Parse v128.const #6275
• We need to make array.new_fixed length annotations mandatory.
  • Make array.new_fixed length annotations mandatory #6277
• We need to remove support for the deprecated text format for string operations.
  • Remove support for legacy stringref text syntax #6289
• The parser (really the lexer) needs to support the non-standard extended name format $"..." and the tests need to be updated to use it where necessary.
  • [Parser] Support string-style identifiers #6278
  • Properly stringify names in tests #6279
• The tests need to be updated to respect the rule that imports must precede declarations.
  • Get more tests working with the new text parser #6284
  • Update lit tests to parse with the new parser #6290
• The parser needs to support struct field names.
  • [Parser] Support references to struct fields by name #6293
• [Outlining] The parser needs to support tuple types. We should also update the format for tuple types to be more consistent with the rest of the language. The parser needs to handle local.set and global.set of tuple types correctly.
  • Update the text syntax for tuple types #6246
  • [Parser] Parse tuple types #6249
  • [Parser] Parse local.set and global.set of tuple values correctly #6250
• [Outlining] The parser needs to support multivalue pops. We should probably update the syntax of pops to be pop <valtype>, where the valtype can be a tuple type.
  • Update pop text syntax #6251
  • [Parser] Parse pops (by doing nothing) #6252
• [Outlining] The parser needs to support pops in control flow structures somehow.
• [Outlining] The parser needs to support the new EH instructions.
  • [Parser] Parse try_table #6237
  • [Parser] Parse throw_ref #6238
• The parser needs to support the implemented typed continuation instructions.
  • [Parser] Parse resume #6295
• The parser (and lexer) needs to support source map comments.
  • [Parser] Parse annotations, including source map comments #6345
  • [Parser] Support prologue and epilogue sourcemap annotations #6370
  • [Parser] Propagate debug locations like the old parser #6377
• The parser match the behavior of the old parser w.r.t unreachable expressions.
  • [Parser] Use unreachables to fulfill tuple requirements #6488
  • [Parser] Pop past unreachables where possible #6489
• The parser should support parsing the full wast scripting language~, including extensions for threads.~
  • [Parser] Parse wast scripts #6581
• We need to either add parser support for Poppy IR or remove the Poppy IR tests.
  • Removed in [Parser] Enable the new text parser by default #6371
• Performance should be better than legacy parser (it is 30% faster after these changes)
  • [Parser][NFC] Solve performance issue by adding maybeLabelidx #6514
  • [Parser] Do not eagerly lex numbers #6544

[kripken]

Perhaps also worth adding #3989 as it blocks us from running the text roundtripping fuzzer, and it would be good to have that fuzzer tested on the new parser:

binaryen/scripts/fuzz_opt.py

Lines 1386 to 1387 in e5948a9

	# FIXME: Re-enable after https://github.com/WebAssembly/binaryen/issues/3989
	# RoundtripText()

[tlively]

The new parser is now on by default, so closing this issue.