Fix matching operator for ignored files in size-labeler.sh

[mhucka]

The IGNORED array in size-labeler.sh contains glob patterns (e.g., *_pb2.py, .*.lock). Using the Bash regex match operator =~ would fail to match these patterns.

The following demonstrates the problem:

# bash -c 'p="*_pb2.py"; [[ a_pb2.py =~ $p ]]'
bash: line 1: [[: invalid regular expression `*_pb2.py': 
Invalid preceding regular expression

This is the same fix applied in the Cirq repo in PR #7986 there.

In addition, this PR adds support for getting more than 100 files back from GitHub, a problem that was flagged by Gemini Code Assist. The solution was implemented with the help of Gemini CLI.

[gemini-code-assist]

Code Review

This pull request updates the size-labeler.sh script to use Bash glob patterns for file exclusion and increases the GitHub API per_page parameter to 100. While the increase in page size is helpful, feedback was provided noting that the script still fails to handle pagination for pull requests exceeding 100 files, which could result in inaccurate size labels.

[mhucka]

/gemini review

[gemini-code-assist]

Code Review

This pull request updates the compute_changes function in dev_tools/ci/size-labeler.sh to implement pagination when querying the GitHub API for pull request files, ensuring accurate size labeling for large changesets. It also improves shell script robustness through better variable quoting and refined pattern matching. Feedback was provided to optimize performance by reducing the number of jq process executions within the file processing loop, suggesting a single-pass approach to extract filenames and change counts.