fix(server): don't raise when tool_use is not followed by tool_result

[Bartok9]

Summary

• validate_tool_use_result_messages() no longer raises a false ValueError when a tool_use block is followed by a plain (non-tool_result) response.
• A model that calls a tool and then replies with text (or the user declining further results) is now accepted, matching SEP-1577.

Motivation

Closes #2960.

Per SEP-1577:

• tool_result blocks MUST be preceded by a tool_use block ✅ (already checked)
• There is no requirement that a tool_use MUST be followed by a tool_result.

The ID-matching block at the end of the function ran whenever the previous message contained a tool_use, regardless of whether the last message actually contained any tool_result blocks:

if has_previous_tool_use and previous_content:
    tool_use_ids = {c.id for c in previous_content if c.type == "tool_use"}
    tool_result_ids = {c.tool_use_id for c in last_content if c.type == "tool_result"}
    if tool_use_ids != tool_result_ids:  # tool_result_ids empty -> mismatch
        raise ValueError(...)

When the last message has no tool_result blocks, tool_result_ids is empty and never equals the non-empty tool_use_ids, so a valid conversation raised ValueError: ids of tool_result blocks and tool_use blocks from previous message do not match.

Fix

Gate the ID-matching check on has_tool_results so it only runs when the last message actually contains tool_result blocks:

if has_tool_results and has_previous_tool_use and previous_content:

This is a one-line change; all existing structural checks (mixed content, missing preceding tool_use, ID mismatch when results ARE present) are unaffected.

Verification

• uv run pytest tests/server/test_validation.py — 16 passed
• uv run ruff check / ruff format --check — clean

Real behavior proof

Regression test test_validate_tool_use_result_messages_no_error_when_tool_use_not_followed_by_tool_result reproduces the exact case from the issue (assistant emits text + tool_use; user replies with plain text). It fails without the fix and passes with it:

# Without the source fix (test present):
tests/server/test_validation.py::...not_followed_by_tool_result  164 / 185  ValueError
Results: 1 failed

# With the fix:
Results: 16 passed