feat(allowlist): add Thrift and Cap'n Proto support

[chethanuk]

Description

.thrift and .capnp files were never reviewed. Neither extension is in supported_file_types.json, so IsAllowedExt drops them at the filter gate before any rule can run.

This adds both to the allowlist and registers **/*.thrift → thrift.md and **/*.capnp → capnp.md in path_rule_map. Two docs rather than one shared IDL doc, because almost nothing carries over between them. Thrift breaks when a required field is added to a live struct, when a retired field id is reused (there is no reserved keyword to hold it open), and when a service method is renamed, since method names travel on the wire. Cap'n Proto breaks when an @N ordinal moves, when a UInt32 is widened to UInt64 (fixed-width slots, unlike a protobuf varint), and when a struct is renamed with no explicit @0x... id pinned, since the id is derived from the name. A merged doc would spend half its lines telling the model which half to ignore. Data only, no Go source changes.

The generated-output excludes are scoped by extension (**/gen-{cpp,cpp2,go,py,...}/**/*.{c,cc,go,py,...}, not a bare **/gen-*/**). IsExcludedPath applies every pattern to every path with no language dispatch, so an unscoped directory glob would drop unrelated files in other languages that happen to sit under a colliding directory name. That is why the file already carries **/Tests/**/*.swift and **/tests/**/*.nim rather than the directory alone. Three negative rows in TestIsExcludedPath fail if anyone broadens it later.

Only the Cap'n Proto C++ header gets a pattern. The generator also emits .capnp.c++, but .c++ was never allowlisted, so those files are already gone by the time IsExcludedPath runs.

Limitation

.thrift and .capnp diffs that used to be skipped are now reviewed. Anyone who does not want that opts out with user exclude globs, which are checked before the allowlist.

Type of Change

• New feature (non-breaking change that adds functionality)

How Has This Been Tested?

• make test passes locally
• Manual testing (describe below)

Rows appended to the existing table tests:

• TestIsAllowedExt — .thrift and .capnp, both cases.
• TestIsExcludedPath — the gen-*, kitex_gen and Cap'n Proto output patterns, plus negatives (tools/gen-go/config.yaml, site/gen-js/index.html, kitex_gen/api/schema.json) that fail if the patterns lose their extension scope, and idl/service.thrift / schema/addressbook.capnp to pin that the schemas themselves stay in review.
• TestResolve_DefaultRules — .thrift and .capnp paths resolve by heading anchors unique to each doc.

Checklist

• My code follows the project's coding style (go fmt, go vet)
• I have performed a self-review of my code
• I have added tests that prove my fix is effective or my feature works
• New and existing unit tests pass locally with my changes
• I have updated the documentation accordingly (if applicable)
• I have signed the CLA

The rule table in pages/src/content/docs/{en,ja,ru,zh}/review-rules.md lists both new patterns.

Related Issues

Part of #470.

[github-actions]

✅ OpenCodeReview: Review complete: 0 finding(s) across 3 selected item(s).

[chethanuk]

@lizhengfeng101 Please review and merge this. Future main updates might cause more conflicts. Thanks :)

[chethanuk]

@lizhengfeng101 Please review and merge this. Future main updates might cause more conflicts. I resolved conflicts 3 times now... Thanks :)

[lizhengfeng101]

@chethanuk I feel that the hit rate for this matching rule is quite low and it is prone to accidental deletion; could we remove it?

Suggested change

"**/gen-{cpp,cpp2,c_glib,cocoa,csharp,netstd,dart,erl,go,hs,hs2,java,js,nodejs,kotlin,lua,perl,php,py,py3,rb,rs,swift}/**/*.{c,cc,cpp,cs,dart,erl,go,h,hrl,hs,java,js,kt,lua,m,mm,php,pl,pm,py,pyi,rb,rs,swift,ts}",

[lizhengfeng101]

LGTM