feat: grammar fixes and add new studio-highlighting.scm file

[hkimura-intersys]

Overview

This PR adds a new studio-highlighting.scm file to group nodes the same way studio does for highlighting purposes. Additionally, fixes to the core grammar were added (see below for details)

Key Changes

1. Parser and scanner behavior were tightened

The core scanner gained explicit handling for inactive preprocessor blocks and a
broader set of command-like tokens.

• common/scanner.h adds POUND_IF_SPECIAL_CASE*, MNEMONIC, and
  TAG_END_IF scanner tokens.
• #if 0 ... #endif blocks are now recognized as a special case so the parser
  can treat unreachable code differently and queries can highlight it as
  inactive. When #IF 0 happens, iris actually registers everything from that line up to #ELSE ,#ENDIF, and #ELSEIF 1 as a comment (so it is technically valid to write whatever you want within those lines). Additionally, here the #ELSE and #ELSEIF 1 happens 100% of the time, so IRIS treats this the same way as an #ENDIF, and makes it so #ENDIF is optional. There were some cases where someone started a block between the #ELSE and #ENDIF, which caused the parser to break. To fix this, in this special case, the #ENDIF is treated as a special tag if #ELSE has already been seen.
• Statement detection (for the objectscript playground grammar) at column 1 was broadened for more command forms and abbreviations, including MV*, ZALLOCATE, ZLOAD, ZUSE, and related
  Z* commands.
• Scanner logic was improved around dotted-line continuations, slash-prefixed
  mnemonics, comments between a condition and a following block, and
  argumentless loop/command termination.

2. Routine Grammar Changes

Before, it was valid to have anything in the first three lines (would return compiled_header token). Now, the second line must be %RO (as all compiled versions have this) to get the compiled_header token. Valid options are compiled_header, routine header (ROUTINE name [type = ]), or an objectscript statement (added this as it is allowed in studio).

• objectscript_routine/grammar.js introduces routine_definition and allows a
  source file to begin with a compiled header, a ROUTINE definition, or a
  top-level statement.
• objectscript_routine/src/scanner.c simplifies the compiled-header and
  ROUTINE detection path and keeps routine-specific lexing active without the
  older staged gating flags.
• rtn_dot handling is now part of the routine grammar extras, which better
  matches real .mac/.rtn content.
• README.md and CONTRIBUTING.md were updated to describe
  objectscript_routine as a routine-file grammar instead of a
  routine-header grammar.

3. Expression, core, and UDL grammars were broadened and cleaned up

Grammar Changes include:

• expr/grammar.js
  • replaces the older pattern-expression regex with a more structured nested
    form (allows ( and ) also)
  • adds $BITLOGIC as a separate built in function, as this one function allows operators that normal expressions don't
  • allows extrinsic indirection with optional method args
  • broadens $SYSTEM.* chaining (as the last item is allowed to be an oref chain expr, not just a regular method)
  • folds parameter-like member access into oref_parameter
• core/grammar.js
  • aligns ZBREAK parsing with the new mnemonic and option tokens
  • allows macros in NEW (since a macro can evaluate to a local var)
  • generalizes MERGE arguments to set_target
  • adds ZLOAD
  • introduces a macro-command form (a macro can be an if statement -> $$$MACRO { w hi} is valid)
  • relaxes some GOTO/argument parsing cases
• udl/grammar.js
  • reuses return_type for property, relationship, and projection typing
  • improves typename parsing for Type(...) Of ...
  • treats more keyword values as type-like names for cleaner trees and
    highlighting
• common/keywords.js and common/identifiers.js were adjusted to support the
  new aliases and identifier behavior.

4. Studio-specific highlighting

• New studio-highlights.scm files were added for:
  • objectscript/queries/
  • objectscript_routine/queries/
  • udl/queries/
• Matching packaged copies were added where needed in Python query directories
  for routine and UDL bindings.
• Rust bindings now export STUDIO_HIGHLIGHTS_QUERY constants in:
  • bindings/rust/lib.rs
  • bindings/rust-routine/lib.rs
  • bindings/rust-playground/lib.rs
• The staged Rust crates now include those query files in their package
  manifests and staging scripts.

5. Standard highlight queries were updated to match the new tree shape

The existing highlights.scm files across expr, core, udl,
objectscript, and objectscript_routine were refreshed.

These changes include:

• parameter_name highlighting moved to oref_parameter
• builtin/system captures were split more cleanly from ordinary identifiers
• new captures were added for zbreak_command_option, tag_end_if,
  command_keyword, keyword_zload, and dotted_statement
• #if 0 regions can now be tagged as inactive code
• UDL/objectscript queries were updated to reflect the return_type-based type
  cleanup

.tsqueryrc.json was updated at the same time to document new capture names
such as:

• comment.inactive
• punctuation.special.dots
• variable.member.oref
• variable.member.sql
• punctuation.bracket.json

6. Query sync and packaging workflows were updated for the new assets

The repo now treats Studio highlight files as related, but separate, artifacts.

• scripts/sync_queries.py still manages the canonical trio:
  • highlights.scm
  • indents.scm
  • injections.scm
• studio-highlights.scm is now explicitly excluded from the normal composed
  query sync and Python query-copy verification path.
• scripts/rust_routine_crate.sh and scripts/rust_playground_crate.sh now
  copy studio-highlights.scm into staged crate directories.
• cargo_readme.md, cargo_readme_playground.md, and
  cargo_readme_routine.md were updated to document the new exported query
  constants.

Testing

• all parsers and queries have been tested and are passing:

This was tested on many real rtn, mac files as well.