perf(pi): connect the pi() sandbox on first use

[eersnington]

Every action on a cold pi() actor opened the session, and opening connected the sandbox. With a remote sandbox, that was a provider round trip for actions that never touch files, and a sandbox outage broke every action.

	Before	After
Any action on a cold actor	connects the sandbox	connects nothing, except to create a new session
A tool call or executeBash	uses the connected sandbox	connects on first use
The sandbox provider is down	every action fails	history and settings actions work; sandbox work fails until the provider is back

• A new session still connects at open, because its working directory comes from the sandbox.
• Tools map paths from the session's working directory onto the sandbox's own (toSandboxPath in src/sandbox.ts), so a replaced sandbox works without changing the session.
• Known gap: after a replacement, an absolute path the model writes inside a bash command still names the old directory until the session reopens.

---

This is part 9 of 11 in a stack:

•  11  fix(pi): persist migrated pi() session entries #330
•  10  feat(pi): expose the pi session to actor code as c.pi #329
•  9  perf(pi): connect the pi() sandbox on first use #328 👈
•  8  feat(pi): drain and resume pi() runs #327
•  7  chore(examples): bump the docs examples to pi-coding-agent 1.0.2 #326
•  6  test(pi): let a run finish and the actor sleep when its caller disconnects #325
•  5  feat(pi): drain pi durable actors across upgrades #324
•  4  feat(pi): stream pi durable conversations to clients #323
•  3  feat(pi): add piDurable actor #322
•  2  feat(pi): store pi durable sessions in actor sqlite #321
•  1  feat(pi): add piDurable() and drain, resume, and c.pi for pi() #320