fix: strip shell comment lines from claude_args before parsing

[dpdanpittman]

shell-quote treats # as a comment character, causing all content after a comment line to be swallowed. This strips lines starting with # before passing to parseShellArgs, so users can add documentation comments to their claude_args without breaking subsequent flags.

Fixes #802

Motivation and Context

As described in #802, users writing multi-line claude_args with explanatory comments like:

claude_args: |
  --model 'claude-haiku-4-5'
  # This is a comment
  --allowed-tools 'mcp__github_inline_comment__create_inline_comment'

find that --allowed-tools is never parsed because shell-quote treats everything after # as a comment object.

Changes

Added a stripShellComments() function that removes lines whose first non-whitespace character is # before passing the input to shell-quote's parse(). This is a line-level filter — hash characters inside quoted values on the same line as a flag are preserved.

File changed: base-action/src/parse-sdk-options.ts

How Has This Been Tested?

Added 4 new tests in base-action/test/parse-sdk-options.test.ts under a new shell comment stripping describe block:

1. Comment line between flags — verifies flags after comments are parsed
2. Indented comment lines — verifies leading whitespace before # is handled
3. Multiple comment lines — verifies consecutive comments don't break parsing
4. Hash inside quoted values — verifies Bash(echo #hello) is preserved

Full test suite passes: 656/656 tests pass.

Breaking Changes

None. This only strips comment lines before parsing. Users who don't use comments see no change.

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)
• Documentation update