[codex] Fix multi-line string tokenization

[nawforce]

Summary

Fixes #27.

Adds explicit support for Summer '26 Apex multi-line string literals that begin with triple single quotes followed by a newline. The outline parser now consumes the literal through the matching closing triple quote instead of treating body apostrophes as regular string delimiters.

Root Cause

consumeStringLiteral handled every single quote as the start or end of a regular Apex string. For triple-quoted literals, an odd number of apostrophes in the body could leave a quote unmatched and make scanning run to EOF.

Changes

• Detect valid triple-quoted openings only when followed by a newline, preserving legacy handling for malformed same-line patterns.
• Consume multi-line string bodies until a matching closing triple quote while preserving existing escape handling.
• Update newline position tracking while consuming multi-line strings so method/body locations remain correct.
• Add smoke tests for balanced bodies, odd apostrophe counts, doubled quotes, unbalanced braces in the string body, and malformed triple-quote fallback.

Validation

• sbt "parserJVM/testOnly io.github.apexdevtools.oparser.SmokeTest"