refactor(client): extract poll_command_completion helper

[devin-ai-integration]

Summary

Three Client::*_and_wait sites duplicated the same ~50-line timeout
loop pattern: check cmd_id.is_ok() for immediate rejection, then
branch between a timeout_ms < 0 infinite poll(50) loop and a
deadline loop driven by a local wait_slice helper, matching terminal
CmdSuccess/CmdError for cmd_id and (for list-style waits)
accumulating data events (BannedUser, UserAccount, ...) along the
way. Every copy agreed today but with small stylistic divergences and
no shared test surface, so any future adjustment (e.g. honouring a new
terminal event, widening the no-op-timeout semantics, reworking how
CommandId::ZERO is reported) would silently drift between call
sites.

Hoist the shape into Client::poll_command_completion next to the
existing poll_until primitive. It is:

• built on top of poll_until (which already handles timeout_ms < 0
  and positive deadlines uniformly);
• parameterised by an error_context string used to rebuild the exact
  same error messages the old copies produced; and
• parameterised by an FnMut(Event, &Message) hook so list-style
  callers can push into a local accumulator without the helper needing
  to know anything about BannedUser, UserAccount, etc.

Applied to:

• Client::wait_for_command (no accumulator; passes |_, _| {}).
• Client::list_bans_and_wait (BannedUser accumulator).
• Client::list_user_accounts_and_wait (UserAccount accumulator).

Deliberately not touched in this PR:

• join_channel_and_wait / login_and_wait - short-circuit on
  ConnectionState changes every iteration, not just on events.
  Folding them in cleanly requires a separate helper shape, or pushing
  the state check into poll_until.
• update_server_and_wait / create_user_account_and_wait /
  delete_user_account_and_wait / wait_for_file_transfer /
  wait_for_file_transfer_terminal - return early on a data event
  rather than on CmdSuccess. Fits a different "first-match-wins"
  helper. Saving that for a follow-up PR to keep this one narrowly
  scoped.

The local wait_slice helpers in channels.rs, server.rs,
directory.rs, files.rs, and users/auth.rs are intentionally kept
as-is. They still have one or more non-converted callers each, and
removing them opportunistically would mask the fact that those sites
are next in the queue.

Structural only, no semantic change:

• Error messages rebuilt verbatim (" rejected in current
  state" / " failed" / Error::Timeout) via
  format!("{error_context} ...").
• cmd_id validation runs before any polling, matching the previous
  early-return behaviour bit-for-bit.
• The poll_until predicate matches exactly the terminal set
  {CmdSuccess, CmdError} tied to cmd_id; non-terminal events flow
  through on_event in the same order as before.
• poll_command_completion is pub(crate) - invisible to downstream
  crates, same surface as the previous private helpers.

Net diff: -81 / +20 lines in existing files, +55 lines in
client/core/runtime.rs (mostly doc comments on the new helper).

Local verification:

• cargo fmt --all --check clean
• cargo clippy --workspace --all-targets --all-features -- -D warnings clean
• cargo test --workspace --all-features -> 287 passed / 0 failed

Review & Testing Checklist for Human

• Skim crates/teamtalk/src/client/core/runtime.rs and confirm the
  new poll_command_completion preserves the original error
  strings bit-for-bit. In particular: "command rejected in current state" / "command failed" for wait_for_command, "ban list command rejected in current state" / "ban list command failed"
  for list_bans_and_wait, "user account list command rejected in current state" / "user account list command failed" for
  list_user_accounts_and_wait.
• Confirm you are happy with the scope: three pure cmd-terminal
  sites converted; the five first-match-wins sites and the two
  state-aborting sites left for follow-up PRs. The alternative is
  one large PR that also introduces a poll_command_or_event
  helper and folds the state-aborting sites in; happy to do that if
  preferred.

Notes

This is PR 2 of the P0 structural refactor queue (PR 1 was #44, which
deduped can_issue_logged_in_command). Next up: either convert the
remaining wait_slice-driven sites to poll_until, or start splitting
the oversized modules (client/bus.rs, bot/storage.rs, bot/fsm.rs,
client/hooks/builders.rs, types/entities/media_common.rs, and the
client/backend*.rs pair) into sub-modules. Each will ship as its own
reviewable PR.

The pre-existing semver CI gate is expected to remain red here too;
release-plz handles the eventual version bump. Every other check is
expected to pass.

Link to Devin session: https://app.devin.ai/sessions/71fdd6196cb74723a2e277bb81993a9c
Requested by: @BlindMaster24

---

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment and CI monitoring

[devin-ai-integration]

✅ Devin Review: No Issues Found

Devin Review analyzed this PR and found no potential bugs to report.

View in Devin Review to see 3 additional findings.