Add Clean Formatting and Markdown paste options

[alvst]

"Paste as Plain Text" is all-or-nothing: it's the only escape from a clip that arrives with a website's fonts and colors attached. This adds two middle options — keep the emphasis, drop the styling, or convert to Markdown.

What changed

• Two new context-menu paste modes alongside Paste as Plain Text, enabled whenever a clip has rich content. Clean Formatting keeps bold, italic, underline, strikethrough, and links while normalizing fonts, sizes, and colors to the system default. Markdown converts rich content to Markdown text, splitting style markers across line breaks so the output stays valid, and rendering links as [text](url).
• The monitor also captures a copy's HTML flavor — browser copies often carry HTML with no RTF — without changing how clips are classified or deduplicated. Conversions prefer HTML, then RTF, then plain text.
• Paste plumbing moves from an asPlainText flag to a PasteFormat enum carrying all four modes.
• The HTML importer is WebKit-backed and will fetch remote subresources during conversion, so every element that can reference an external resource (img, picture, source, iframe, object, embed, link, script, style, svg, video, audio, and CSS url()) is stripped before conversion, backed by a 2-second importer timeout. Sources over 1 MB skip rich conversion and fall back to plain text.
• Capture-side, stored pasteboard HTML is capped at the sync layer's own inline payload bound (CKSchema.inlineLimit); oversized HTML is dropped and conversions fall back to RTF/plain text as before.

Screenshots

Before:

After:

Notes for review

• htmlData is not carried by iCloud sync, so the same clip can produce different Markdown on a synced Mac — the one that captured it converts from HTML, the other falls back to RTF or plain text. It degrades gracefully rather than failing, and extending the sync schema for it felt like the wrong thing to bundle into this PR. Stating it rather than letting it be discovered.
• Markdown scope is deliberately narrow: text plus bold, italic, underline, and links. Headings, lists, and strikethrough are not emitted — Clean Formatting keeps strikethrough, Markdown doesn't. The converters only round-trip what they can represent faithfully; guessing at heading levels from font sizes produces confidently wrong output.
• The strip-before-convert step means "Paste as Markdown" never issues a network request at paste time — a tracking pixel in a captured mail or web clip would otherwise have been fetched synchronously on the main thread.
• Concealed clips bail before HTML is ever read, and htmlData inherits the store's 0600 permissions.
• No dependencies; merges in any order.

---

This feature should be bundled into v2.

Part of #80.

[momenbasel]

Landed as 479bacd through #104 with a follow-up (c7cdf8a). The regex pass that ran before the HTML importer rewrote visible text along with markup: a url(...) in the clip's own words, a copied CSS snippet or the URL("https://...") call in a code sample pasted as "none" through Clean Formatting and Markdown. It also guarded against something NSAttributedString(html:) does not do; a local HTTP server logged no request for an unsanitized img, input type=image, body background, image-set() or meta refresh. The HTML now reaches the importer as captured, with the 2 s document timeout and the 1 MB conversion cap kept.

Two more things the review turned up: editing a clip replaced its text and RTF but kept the captured HTML, which the converters prefer, so Paste as Markdown after an edit produced the pre-edit words (both edit paths clear it now), and HTML is captured only when the pasteboard carries no RTF and only up to 64 KB, since store.json is re-encoded after every capture and Safari copies already carry RTF. Thanks @alvst.