Improve Go comments filtering

[Vladyslav-Kuksiuk]

This PR improves Go comments filtering.

Resolves this issue.

[Oleg-Melnik]

Claude Review

LGTM — nothing blocking found.

Checked and confirmed clean:

• The Escapes distinction is semantically right: Go raw strings have no escapes, so the first backtick always closes — the old shared textBlockEnd would have mis-skipped on `C:\Users\` (the backslash-skip consumed the closing delimiter). Java text blocks do honor \", preserved via Escapes: true. Both now match their language specs.
• Scan order stays safe: consumeTextBlockStart runs before consumeQuotedSegment, so Java's """ still wins over ""+" disambiguation, and Go's backtick no longer collides since it was removed from goQuoteChars.
• Multi-line state: textBlockActive/textBlock persist across lines via blockState; unterminated-at-EOL writes the remainder and continues on the next line — the new multi-line test exercises exactly this.
• Removed behavior: dropping the backtick from Go's quote chars removes the old (incorrect) single-line escape-processing path — its replacement is strictly better. No syntax besides Java and Go used TextBlocks, so the []string → []TextBlockMarker change is fully covered.
• Edge cases hold: trailing backslash at end of line in an escaping block (cursor overshoot exits the loop cleanly, state stays active); escaped quotes before potential delimiters; "" adjacent to text; prefixFrom is still live via prefixAt (no dead code).

One micro-nit, take or leave: goQuoteChars = "\"'" duplicates the literal Java already uses inline for its QuoteChars — a single shared const (e.g. cLikeQuoteChars) would cover both. Purely cosmetic.

[Oleg-Melnik]

@Vladyslav-Kuksiuk LGTM with minor comment from Claude.