fix(envd): skip MMDS polling on /init when running outside Firecracker

[Luckydog691]

Problem

envd started with -isnotfc runs outside Firecracker. The flag is documented as "run outside of Firecracker (skips MMDS poll and HTTP log exporter)", and main.go correctly guards its startup poll with if !isNotFC.

The /init handler, however, starts the same 60s MMDS polling loop unconditionally. Every attempt opens a TCP connection to 169.254.169.254:80; with no MMDS endpoint reachable, the handshake never completes and the connection stays in SYN_SENT for the whole poll window (visible in /proc/net/tcp).

That dangling half-open connection becomes fatal when the sandbox is checkpointed: a gVisor (runsc) checkpoint taken while such a connection exists panics in the TCP endpoint save path (resetConnectionLocked dereferences a nil snd/rcv), so checkpointing fails deterministically for the whole poll window.

Fix

Mirror the guard main.go already applies to its startup poll and skip the MMDS polling in /init when isNotFC is set. Firecracker-based deployments are unchanged.

Testing

• In a containerized deployment, a SYN_SENT connection to 169.254.169.254:80 is observable for ~60s after /init before the change; with the change it never appears.
• Checkpointing a runsc sandbox that is inside that window fails before the change and succeeds after it (checkpoint + restore round trip verified).
• go build ./... and go test ./internal/api/ pass.

[cla-bot]

We require contributors to sign our Contributor License Agreement, and we don't have @Luckydog691 on file. You can sign our CLA at https://e2b.dev/docs/cla . Once you've signed, post a comment here that says '@cla-bot check'