storagefsm: Fix unsealedInfoMap.lk init race

[magik6k]

This:

goroutine 38357 [semacquire, 17 minutes]:
sync.runtime_SemacquireMutex(0xc01089f264, 0xc0001b9700, 0x1)
	/usr/local/go/src/runtime/sema.go:71 +0x47
sync.(*Mutex).lockSlow(0xc01089f260)
	/usr/local/go/src/sync/mutex.go:138 +0x105
sync.(*Mutex).Lock(...)
	/usr/local/go/src/sync/mutex.go:81
github.com/filecoin-project/lotus/extern/storage-sealing.(*Sealing).restartSectors(0xc01089f1d0, 0x32c6520, 0xc01279c700, 0x0, 0x0)
	/opt/lotus/extern/storage-sealing/fsm.go:412 +0x8ea
github.com/filecoin-project/lotus/extern/storage-sealing.(*Sealing).Run(0xc01089f1d0, 0x32c6520, 0xc01279c700, 0xc012312990, 0xc01278f814)
	/opt/lotus/extern/storage-sealing/sealing.go:152 +0x5a
created by github.com/filecoin-project/lotus/storage.(*Miner).Run
	/opt/lotus/storage/miner.go:164 +0x589

Could happen when there were a bunch of deals pending on restart, and the market sent them to storage-fsm before it started initializing, leading to a deadlock