Implement Ignore Pattern Support

[tim-dickey]

Goal

Allow users to exclude files and/or content regions (e.g., license headers, generated code) from duplicate scanning.

Scope

• CLI flags: --ignore-glob (repeatable), --ignore-file PATTERN_FILE.
• Pattern file format: one glob per line, # comments, blank lines ignored.
• Content region exclusion: configurable regex patterns applied before tokenizing (e.g., --ignore-region 're:^# Generated.*?\n').
• Update DuplicateFinder to apply filters pre-tokenization.

Acceptance Criteria

• Files matching ignore globs are skipped (not counted in scanned set).
• Region regex patterns remove matched text before tokenization.
• Tests: file exclusion, region removal reduces shingles, no false negatives for remaining content.
• README section documenting patterns + examples.

Non-Goals

• Language-aware comment stripping (future tokenizer feature).

Future

• Central config file (duplicate-finder.toml) for ignore patterns.