CI: randomly failed e2e test and continuous prefetch warnings

[hangvane]

https://github.com/containerd/nydus-snapshotter/actions/runs/6739821947/job/18321972302

time="2023-11-03T00:41:46Z" level=fatal msg="connection error: desc = \"transport: Error while dialing dial unix:///run/containerd-nydus/containerd-nydus-grpc.sock: timeout\": unavailable"
[2023-11-03 00:41:46.650093 +00:00] WARN [storage/src/cache/worker.rs:364] storage: failed to prefetch data from blob e4d6efbd0a0d130541b68d7fe421474184657b37eff8fef1d6e3048454b0c671, offset 0, size 6424, Invalid argument (os error 22): storage/src/meta/mod.rs:568, will try resend
[2023-11-03 00:41:46.650321 +00:00] WARN [storage/src/cache/worker.rs:364] storage: failed to prefetch data from blob b194a89b4d14dfe137e08bc984b7491b079e169597fab03c6e92d1c7e76fbc97, offset 0, size 5697, Invalid argument (os error 22): storage/src/meta/mod.rs:568, will try resend
[2023-11-03 00:41:46.650374 +00:00] WARN [storage/src/cache/worker.rs:364] storage: failed to prefetch data from blob 35abf43dce465d40f45a4445af0c570dcd4dc36faa6a225928aeb64c122c3a08, offset 10485760, size 146731, Invalid argument (os error 22): storage/src/meta/mod.rs:568, will try resend
[2023-11-03 00:41:46.651170 +00:00] WARN [storage/src/cache/worker.rs:364] storage: failed to prefetch data from blob a8f84a5d40140aa0586b41668a28f9579d0a6c9dc0fdc8abb09197cef00a2cef, offset 0, size 9415, Invalid argument (os error 22): storage/src/meta/mod.rs:568, will try resend

Seems that the prefetch warnings in e2e test exist for at least 2 months, and may randomly fail e2e test.

Additionally, the warning infomation is covered by Invalid argument (os error 22), making it hard to trace the error.

[imeoer]

It seems this error happened earlier:

time="2023-11-03T00:41:42.464732357Z" level=warning msg="snapshot garbage collection failed" error="error reading from server: EOF: unavailable" snapshotter=nydus

It looks like the nydusd HTTP API server is broken.