fix(linter/plugins): handle zero-token files in SourceCode#getLastToken()

[lilnasy]

• The root cause of the "iffy" line of code in feat(linter/plugins): implement SourceCode#getLastToken() #16003 was an incorrectly implemented test.
• The corrected test revealed that the code attempted to index the 0 index of an empty array.
• This PR fixes both issues.
• Also took the opportunity to remove a variable declaration in the binary search.

[Copilot]

Pull request overview

This PR fixes a bug in getLastToken() where it incorrectly handled zero-token files (e.g., files containing only comments). The binary search algorithm was refactored to properly handle empty token arrays, and the tests were corrected to use accurate ranges matching the actual source text length.

• Refactored binary search to initialize lastTokenIndex to 0 and decrement after the loop, eliminating the "iffy" early return
• Fixed test ranges to match actual source text lengths (e.g., "// comment" is 10 chars, not 11)
• Removed unnecessary lo variable from binary search by using lastTokenIndex as the lower bound

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
apps/oxlint/src-js/plugins/tokens.ts	Refactored binary search to handle zero-token arrays correctly by initializing lastTokenIndex to 0 and decrementing after the loop, removing the conditional early return
apps/oxlint/test/tokens.test.ts	Corrected test ranges from incorrect values (using Program range or wrong lengths) to accurate ranges matching the actual source text

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.