fix(fetchkit): tighten content-type checks for markdown and text

[chaliy]

Motivation

• Substring-based Content-Type checks could misclassify text/html responses that include parameters like profile="text/markdown" as markdown or plain text, allowing raw HTML (including <script>/iframe) to bypass the HTML conversion sanitization.

Description

• Change is_markdown_content_type and is_plain_text_content_type to parse only the primary media type token (the part before ;) and compare it with eq_ignore_ascii_case after trimming.
• Add regression assertions to the unit tests to ensure headers like text/html; profile="text/markdown" and text/html; profile="text/plain" are not treated as markdown/plain text.
• This is a minimal behavioral fix that preserves existing functionality for legitimate text/markdown and text/plain media types while preventing parameter-based spoofing.

Testing

• Ran cargo test -p fetchkit is_markdown_content_type and the updated test_is_markdown_content_type passed.
• Ran cargo test -p fetchkit is_plain_text_content_type and the updated test_is_plain_text_content_type passed.
• The change is limited to content-type detection and does not alter the HTML conversion code paths beyond restoring correct sanitization routing.

---

Codex Task