fix: rich messages follow-ups — reply_parameters, send latch, opt-in default

- Use reply_parameters per the sendRichMessage spec instead of the
  undocumented reply_to_message_id scalar (silently ignored -> reply
  anchor quietly dropped).
- Latch rich sends off after an endpoint-capability failure (old PTB /
  server without sendRichMessage) so every later reply doesn't pay a
  doomed extra roundtrip; per-message BadRequests do NOT latch.
- Default rich_messages to OFF (opt-in) while the day-old Bot API 10.1
  endpoint is validated live; revert the prompt-hint table guidance
  until the default flips on.
- Tests: reply_parameters shape, send-latch behavior, BadRequest
  non-latch; rich tests opt in explicitly via extra.

tests/agent/test_prompt_builder.py

@@ -877,20 +877,6 @@ class TestPromptBuilderConstants:
         # check that this test is calibrated correctly).
         assert "include MEDIA:" in PLATFORM_HINTS["telegram"]
 
-    def test_telegram_hint_encourages_rich_markdown(self):
-        # Regression: Telegram now supports Bot API 10.1 Rich Messages, so the
-        # hint must encourage tables / task lists / rich Markdown and must no
-        # longer forbid tables. The adapter sends final replies via
-        # sendRichMessage with raw markdown (see test_telegram_rich_messages).
-        hint = PLATFORM_HINTS["telegram"]
-        lowered = hint.lower()
-        assert "Telegram has NO table syntax" not in hint
-        assert "table" in lowered
-        assert "task list" in lowered
-        assert "rich markdown" in lowered
-        # Local media delivery guidance must remain intact.
-        assert "include MEDIA:" in hint
-
     def test_platform_hints_mattermost(self):
         hint = PLATFORM_HINTS["mattermost"]
         assert "Mattermost" in hint