fix(simplex): extract group sender from chatDir.groupMember

[brandon-btcgroup]

What does this PR do?

Fixes group-message sender extraction in the SimpleX Chat platform adapter
(plugins/platforms/simplex/adapter.py).

For group messages, current simplex-chat reports the sending member under
chatItem.chatDir.groupMember. The adapter only read the legacy
chatItem.chatItemMember key, which is absent in this payload shape. As a
result sender_id fell back to the chat_id (e.g. "group:1") — which is
never a real member id — and downstream allowlist matching fails, so the
bot can't identify who sent a group message even when delivery otherwise works.

The fix reads chatDir.groupMember first and falls back to chatItemMember
for older daemon payloads, and resolves the display name via memberProfile
when the member object doesn't carry it directly.

Related Issue

Related to #30150, which documents three other SimpleX adapter bugs (missing
/_start subscribe, newChatItems resp-nesting, and /_send command syntax).
This PR addresses a fourth, independent bug — sender extraction — that is
not tracked by that issue. Happy to open a dedicated issue if maintainers
prefer one on file.

Fixes #35045

Type of Change

• 🐛 Bug fix (non-breaking change that fixes an issue)

Changes Made

• plugins/platforms/simplex/adapter.py — read the group sender from
  chatItem.chatDir.groupMember, falling back to the legacy
  chatItem.chatItemMember; resolve displayName via memberProfile.
• tests/gateway/test_simplex_plugin.py — add
  test_group_sender_from_chatdir_groupmember asserting the correct
  memberId/display name are extracted from the chatDir.groupMember layout.

How to Test

1. Run a simplex-chat daemon and join the bot to a group.
2. Send a group message from a phone and observe the inbound event: the member
   lives under chatItem.chatDir.groupMember (no chatItemMember key present).
3. Before: sender_id resolves to the chat_id ("group:<n>"), so the
   member is unidentifiable and allowlist checks against the real memberId
   fail. After: sender_id is the member's base64 memberId and the
   display name resolves correctly.
4. Unit test: scripts/run_tests.sh tests/gateway/test_simplex_plugin.py
   (28 passing here, incl. the new regression test).

Note: reproducing the full phone→reply round-trip on a vanilla daemon also
requires the three fixes in #26433 (without /_start, no events arrive at
all). The added unit test exercises the sender-parsing path in isolation, so
it stands on its own. Verified end-to-end on a live Ubuntu 24.04 deployment
with the #26433 fixes applied — correct memberId + display name extracted.

Relationship to existing PRs (duplicate check)

I searched open PRs before submitting:

• Fix(gateway): 3 bugs blocking bidirectional simpleX messaging #26433 (focused, open) fixes the three bugs in hermes-simplex-bugs-report #30150 — but not this
  sender-extraction bug. This PR is complementary, not a duplicate; it isolates
  the one fix that's missing from that PR.
• feat(simplex): add SimpleX Chat platform adapter #4666 and feat(simplex): groups, native attachments, text batching, auto-accept #27978 (large feature PRs, open) do include this same
  chatDir.groupMember change, but bundled inside hundreds of lines of other
  work. This PR offers it as a small, independently-reviewable fix.

If maintainers would rather fold this single change into #26433 or one of the
feature PRs, I'm happy to close this and move the diff there.

Checklist

Code

• I've read the Contributing Guide
• My commit messages follow Conventional Commits (fix(simplex): …)
• I searched for existing PRs to make sure this isn't a duplicate (see above)
• My PR contains only changes related to this fix (one commit, two files)
• I've run the test suite via scripts/run_tests.sh — SimpleX plugin suite 28/28 pass. The full suite has some failures, but they are pre-existing on unmodified main (anthropic adapter, gateway service/WSL/systemd, TUI — environment-specific on macOS) and confirmed unrelated to this change; this PR introduces no new failures.
• I've added tests for my changes
• I've tested on my platform: Ubuntu 24.04 (live deployment) + macOS (unit tests)

Documentation & Housekeeping

• Docs — N/A (internal parsing fix, no user-facing surface)
• cli-config.yaml.example — N/A (no config keys changed)
• CONTRIBUTING.md / AGENTS.md — N/A
• Cross-platform impact considered — N/A (pure dict parsing)
• Tool descriptions/schemas — N/A

[teknium1]

Thanks for isolating the sender-parsing compatibility case.

Problems

• The diff targets _handle_new_chat_item, but current main moved inbound parsing to _handle_chat_item (plugins/platforms/simplex/adapter.py:472) in 0c2e81df0. The proposed implementation and its direct test call therefore need to be transplanted rather than applied verbatim.
• Current main already implements the test's chatDir.groupMember + memberProfile.displayName case at plugins/platforms/simplex/adapter.py:526-530. The remaining contribution is the legacy chatItemMember fallback, but the PR adds no regression coverage for that path.

Suggested changes

• Apply the fallback in the current group branch beginning at plugins/platforms/simplex/adapter.py:526.
• Test it through _handle_chat_item using a received-message payload and configured group allowlist, asserting the legacy member object supplies both source ID and display name.

Automated hermes-sweeper review.