[Investigate] Add a post-provision MCP-reload nudge to pod.py provision

Parent: #275 (audit item 1)

Background

Per the audit posted on #275 (item 1), there is a gap between pod.py provision writing the new pod to ~/.claude/mcp.json and Claude Code's MCP runtime picking up the change. The user must manually /mcp to reload before SSH MCP tools resolve for the freshly-provisioned pod. The lifecycle code IS correct — pod_config.update_mcp_config writes the right file at pod_config.py:275-340 — the gap is in the runtime reload flow.

In production this is a non-issue because the /issue skill exits after Step 6b and the experimenter starts in a NEW Claude Code session where MCP loads fresh. The gap only bites when an agent provisions a pod and then tries to immediately SSH into it without an intervening session boundary.

Two viable fixes

1. Cosmetic. Emit a → Run /mcp in Claude Code to reload SSH MCP tools for this pod line at the end of pod.py provision output. Zero risk.
2. Functional. Add a one-shot pod.py mcp-reload command that invokes Claude Code's reload via the claude CLI if available. Higher risk — depends on the claude CLI's reload semantics.

Out of scope

Pre-loading SSH MCP tools at provision-time inside the same Claude Code session — that's a Claude Code runtime feature, not a project-side bug.