feat(telegram): convert entities to markdown in parsed messages

[dzhng]

Summary

Telegram delivers formatting (bold, italic, links, code, etc.) as separate entity objects alongside plain text. Previously, parseTelegramMessage only used the raw text field, losing all entity information — most critically, text_link entity URLs were dropped entirely.

For example, a Telegram message with "Visit our website for details" would arrive as:

• text: "Visit our website for details"
• entities: [{ type: "text_link", offset: 10, length: 7, url: "https://example.com" }]

The link URL was completely lost in the parsed message.text.

Changes

• Adds applyTelegramEntities() which reconstructs markdown syntax from Telegram entities before storing the message text
• Supported entity types: text_link → [text](url), bold → **text**, italic → *text*, code → `text`, pre → ```text```, strikethrough → ~~text~~
• Other entity types (url, mention, bot_command) are already present in the raw text and left unchanged
• Adds missing url and language fields to TelegramMessageEntity type
• Passes original plain text to isBotMentioned() to preserve correct entity offset calculations

Test plan

• 12 new unit tests for applyTelegramEntities covering all entity types, edge cases, and integration with parseMessage
• All 101 existing tests pass
• Lint/format clean (pnpm -w run check passes)

[vercel]

@dzhng is attempting to deploy a commit to the Vercel Labs Team on Vercel.

A member of the Team first needs to authorize it.