fix(slack): render <@user>, <@subteam^group> and <#channel> mentions as rich-text elements

[CumpsD]

The Block Kit converter parses bold/italic/code/links but passed Slack-native mention syntax (<@U...>, <@subteam^S...>, <#C...>) through as literal text: rich_text blocks render their own elements and ignore the mrkdwn Text fallback on modern clients, so agents had no way to mention users in outbound Slack messages.

Add mention patterns to the inline parser emitting RichTextUser, RichTextUserGroup and RichTextChannel (all present in SlackNet 0.17.x), including the labeled forms (<@U123|name>, <#C123|general>). Covers paragraphs, list items and quotes via the shared ParseInlineElements.

Adds 5 converter tests: user, labeled user, user group, channel, and user mention inside a bullet list item. All 30 SlackBlockConverterTests pass locally.

[Aaronontheweb]

Reviewed the diff. The user and channel mention paths look correct and well-tested. The usergroup path has a bug and a gap:

1. usergroup_id includes the subteam^ prefix (high)

UserGroupMentionRegex is <@(subteam\^[S0-9A-Z]+)> and the handler assigns m.Groups[1].Value — the full token subteam^S0123ABC — to RichTextUserGroup.UserGroupId. Slack's rich-text usergroup element expects the bare group ID (e.g. S0615G0KT from usergroups.list); a subteam^-prefixed value won't resolve, so the mention still renders as literal text — the exact failure this PR is fixing, for the one usergroup shape it matches. The new test asserts "subteam^S0123ABC", so the bug is baked into the test.

Suggested fix: <@subteam\^([S0-9A-Z]+)> and assert "S0123ABC".

2. Labeled usergroup form not matched (medium)

<@subteam^S0123ABC|@eng-team> — the common labeled form — doesn't match any of the new regexes (the user regex fails on lowercase s, the usergroup regex has no optional label), so it renders as literal text. The user and channel regexes both accept (?:\|[^>]+)? labels; suggest giving the usergroup regex the same treatment: <@subteam\^([S0-9A-Z]+)(?:\|[^>]+)?>.

3. Channel regex covers C-prefixed IDs only (low)

<#G123|private> (private channels, group DMs) stays literal since the regex is C[0-9A-Z]+. This matches Slack's documented #C form, but real workspace links to private channels won't convert. Consider [CGD][0-9A-Z]+.

Nits: the BOM removal on both files is harmless noise.

Verdict: changes requested. User and channel paths are solid; the usergroup path needs the ID fix before it's functional.

[CumpsD]

Addressed all review remarks in 3766eea: bare usergroup ID capture (test assertion updated too), labeled usergroup form, [CGD] channel prefixes, and BOMs restored. 32/32 SlackBlockConverterTests pass.

What changed per remark:

1. High — UserGroupMentionRegex now captures the bare ID (S0123ABC), test asserts the same
2. Medium — usergroup regex accepts the labeled form <@subteam^S123|oncall>, new test added
3. Low — channel regex accepts C/G/D prefixes (private channels, group DMs), new test added
4. Nit — UTF-8 BOMs restored on both files

[Aaronontheweb]

Thanks for the contribution @CumpsD! We've folded your commit into #1763, which includes your original commit unchanged plus follow-up fixes from review (usergroup_id now uses the bare group ID, plus support for the labeled and <!subteam^ bang forms). Auto-merge is enabled on #1763 — closing this one in favor of it. Credit for the feature is all yours. 🎉

[CumpsD]

👍🏻

[Proxicon]

Cool. Wonder if this works in discord.
Will need to test!