feat(mcp): batch read tools per high-traffic vertical

[dougborg]

Problem

Real Front workflows often fetch multiple resources in sequence: "summarize the last 5 messages from each of these 3 conversations", "look up these 10 customer handles", "tag these 6 tickets". Each per-id round trip burns LLM context and API quota.

Sibling repo statuspro added batch read tools in PR #49 and saw meaningful reductions in tool-call count for multi-item workflows.

Solution

Add 2-3 batch read tools per high-traffic vertical. Initial set:

• get_conversations_batch(ids: list[str]) - fetch up to N conversations in one call
• get_messages_batch(ids: list[str]) - fetch up to N messages in one call
• lookup_contacts_batch(handles: list[str]) - resolve handles to contacts in one call

Implementation:

• Parallel fan-out via asyncio.gather (transport-layer rate limiting already handles concurrency safely)
• Cap at a sensible per-call max (~25) to keep responses bounded
• Return aggregated results with per-id error reporting (a single failed lookup must not sink the batch)

Files to change

• frontapp_mcp_server/src/frontapp_mcp/tools/conversations.py
• frontapp_mcp_server/src/frontapp_mcp/tools/messages.py
• frontapp_mcp_server/src/frontapp_mcp/tools/contacts.py
• frontapp_mcp_server/src/frontapp_mcp/resources/help.py
• frontapp_mcp_server/tests/tools/

Acceptance criteria

• All three batch tools implemented and registered
• Each tool caps batch size and returns clear error per failed id
• Tests cover happy path, max-size, and partial-failure
• Help resource updated

Reference

• feat(mcp): add batch read tools (get_orders_batch, lookup_orders_batch, summarize_active_orders) statuspro-openapi-client#49