fix: correct tool repetition detector to not block first tool call when limit is 1

[roomote]

This PR fixes issue #6834 where setting the "Errors and Repetition Limit" to 1 incorrectly blocks the first (non-repeated) tool call.

Problem

The tool repetition detector was incorrectly counting the first occurrence of a tool as 1, which meant that with a limit of 1, the very first tool call would be blocked even though it was not a repetition.

Solution

• Changed the initial count from 1 to 0 for new (non-repeated) tool calls in ToolRepetitionDetector.ts
• Updated all affected test cases to match the correct expected behavior
• Now a limit of N correctly allows N consecutive identical calls before blocking (as intended)

Testing

• All existing tests have been updated and pass
• The fix ensures that:
  • Limit of 1: Allows the first call, blocks the second identical call
  • Limit of 2: Allows first two calls, blocks the third identical call
  • And so on...

Related Issue

Fixes #6834

---

Important

Fixes tool repetition detection to correctly allow the first tool call when the limit is set to 1, updating logic and tests accordingly.

• Behavior:
  • Fixes issue Errors and Repeated Use Limitations #6834 by changing initial count from 1 to 0 for new tool calls in ToolRepetitionDetector.ts.
  • Ensures a limit of N allows N consecutive identical calls before blocking.
• Testing:
  • Updates test cases in ToolRepetitionDetector.spec.ts to reflect the corrected behavior.
  • Tests confirm that:
    • Limit of 1 allows the first call, blocks the second.
    • Limit of 2 allows the first two calls, blocks the third.
    • Similar logic applies for higher limits.

^{This description was created by for 72691ac. You can customize this summary. It will automatically update as commits are pushed.}

[roomote]

Reviewing my own code is like debugging in a mirror - everything looks backwards but the bugs are still mine.

[roomote]

Good fix! The change from 1 to 0 is correct - the first occurrence shouldn't count as a repetition. Consider adding a comment here to explain the counting logic for future maintainers, something like: // Start from 0 since this is the first occurrence, not a repetition

[daniel-lxs]

LGTM