fix(agent): strip finish_reason from assistant messages to fix Mistral 422 errors

[Sertug17]

Problem

Fixes #134

When using Mistral API directly (api.mistral.ai), the second message always fails with HTTP 422:

'loc': ['body', 'messages', 2, 'assistant', 'finish_reason'],
'msg': 'Extra inputs are not permitted'

Root Cause

In _build_assistant_message(), a finish_reason field is stored on every assistant message dict for internal trajectory tracking. When these messages are replayed as conversation history in subsequent API calls, the finish_reason field is sent to the API.

Mistral's API strictly forbids extra fields in message objects and returns 422. The first message always works because there's no history yet — the error appears from the second message onward.

Fix

Strip finish_reason and reasoning from assistant messages before sending them to the API, in both the main agent loop and the memory flush loop:

if "reasoning" in api_msg:
    api_msg.pop("reasoning")
# Remove finish_reason - not accepted by strict APIs (e.g. Mistral)
if "finish_reason" in api_msg:
    api_msg.pop("finish_reason")

Testing

Verified that finish_reason is no longer present in outbound API message payloads.

[Sertug17]

Note: While other PRs (#138, #250) address the finish_reason stripping issue, this PR also includes a fix for the reasoning parameter being sent to Mistral's API, which independently causes 422 errors. The reasoning field in extra_body is only valid for OpenRouter/Nous Portal endpoints — sending it directly to api.mistral.ai results in an "Extra inputs are not permitted" error regardless of the message history fix.
These two fixes together provide complete Mistral API compatibility.

[teknium1]

LGTM