Handle independent OSC8 hyperlink sequences

[Spidy092]

Fixes #273.

Colored text inside an OSC8 hyperlink leaves control sequences in the text used to calculate table widths. For example, a hyperlink containing a red red followed by link should occupy eight columns, but the existing parser counts the hyperlink controls when wide-character handling is disabled. It also misses valid direct transitions between hyperlinks and standalone closing sequences.

Recognize each OSC8 control sequence independently and strip only the controls, preserving the displayed text. Update the word wrapper to consume complete control sequences and keep hyperlink state separate from color state; close hyperlinks at cell boundaries and reopen them on continuation lines.

Add regression coverage for colored link text, direct URI changes, standalone opening and closing controls, link IDs containing punctuation, string and byte inputs, table alignment, and wrapped hyperlink text. Update the parser comments and stripping/wrapping docstrings to describe the behavior.

Validation on Linux with Python 3.14.4:

• Unchanged baseline: 382 tests passed, 1 skipped.
• Before implementation, the initial regression tests produced 11 failures and 1 pass.
• Final full suite, including doctests, NumPy, Pandas, and wcwidth: 398 passed, 1 skipped.
• Ruff 0.15.4, as pinned in the project's pre-commit configuration: lint and formatting passed.
• Source distribution and wheel build passed with python3 -m build --no-isolation.
• git diff --check passed.

Tests ran offline in an isolated filesystem without GitHub credentials. Other Python versions and operating systems have not been tested locally; upstream CI has not run. The build used a shallow checkout, so its generated development version is not release metadata.

Prepared with Codex assistance.