Improve C# comments filtering

[Vladyslav-Kuksiuk]

This PR improves the C# comments filtering.

Resolves this issue.

[Oleg-Melnik]

Claude Review

Three non-blocking findings:

1. C# 11 raw string literals (""" / $""") are unhandled. A multi-line """ block degrades into empty-string + open-string parsing, so //-like text inside a raw string gets stripped as a comment — silent content corruption in embedded snippets. Probably a follow-up issue rather than a blocker for this PR.
2. /*/ self-closes. consumeComment doesn't advance past the /* opener, so consumeActiveBlock's strings.Index search finds the overlapping */ at offset 1 and treats /*/ as a complete comment, leaking the actual comment body as code. Inherited from the identical pattern in the JS and marker filters, so a shared fix would cover all three.
3. Duplication. consumeActiveBlock, consumeCodeByte, writeEscapedByte, and the Filter loop skeleton are now byte-identical third copies of the JavaScript filter's versions. A shared line-scanner core would keep the three lexers from drifting.

[Oleg-Melnik]

@Vladyslav-Kuksiuk LGTM with Claude comments to be addressed before merging.