fix(share/availability): add missing protection for autobatch (#3988)

Initially reported as panic by @Wondertan 
```
fatal error: concurrent map iteration and map write goroutine 424607 [running]: [github.com/ipfs/go-datastore/autobatch.(*Datastore).Flush(0xc0015ff300](http://github.com/ipfs/go-datastore/autobatch.(*Datastore).Flush(0xc0015ff300), {0x3aacc20, 0xc00bb3b340}) /go/pkg/mod/github.com/ipfs/[email protected]/autobatch/autobatch.go:108 +0xa5 [github.com/ipfs/go-datastore/autobatch.(*Datastore).Put(...)](http://github.com/ipfs/go-datastore/autobatch.(*Datastore).Put(...)) /go/pkg/mod/github.com/ipfs/[email protected]/autobatch/autobatch.go:67 [github.com/celestiaorg/celestia-node/share/availability/light.(*ShareAvailability).SharesAvailable(0xc00050ba40](http://github.com/celestiaorg/celestia-node/share/availability/light.(*ShareAvailability).SharesAvailable(0xc00050ba40), {0x3aacc20, 0xc00bb3b340}, 0xc00694c180) /src/share/availability/light/availability.go:154 +0xab9 [github.com/celestiaorg/celestia-node/das.(*DASer).sample(0xc0014da1a0](http://github.com/celestiaorg/celestia-node/das.(*DASer).sample(0xc0014da1a0), {0x3aacc20, 0xc00bb3b340}, 0xc00694c180) /src/das/daser.go:171 +0xc9 [github.com/celestiaorg/celestia-node/das.(*worker).sample(0xc00bdb6480](http://github.com/celestiaorg/celestia-node/das.(*worker).sample(0xc00bdb6480), {0x3aacbb0, 0xc002286a00}, 0x2540be4000, 0xc00fa7feff?) /src/das/worker.go:121 +0xe2 [github.com/celestiaorg/celestia-node/das.(*worker).run(0xc00bdb6480](http://github.com/celestiaorg/celestia-node/das.(*worker).run(0xc00bdb6480), {0x3aacbb0, 0xc002286a00}, 0x2540be4000, 0xc0018fa3f0) /src/das/worker.go:81 +0x254 [github.com/celestiaorg/celestia-node/das.(*samplingCoordinator).runWorker.func1()](http://github.com/celestiaorg/celestia-node/das.(*samplingCoordinator).runWorker.func1()) /src/das/coordinator.go:112 +0x68 created by [github.com/celestiaorg/celestia-node/das.(*samplingCoordinator).runWorker](http://github.com/celestiaorg/celestia-node/das.(*samplingCoordinator).runWorker) in goroutine 560 /src/das/coordinator.go:110 +0x317
```

The panic was caused by concurrent access to autobatch datastore due to lack of sync protection on Flush call. This PR adds missing mutex protection for Flush call, as well as Delete, which also had protection missing .

share/availability/light/availability.go

@@ -222,7 +222,9 @@ func (la *ShareAvailability) Prune(ctx context.Context, h *header.ExtendedHeader
 	}
 
 	// delete the sampling result
+	la.dsLk.Lock()
 	err = la.ds.Delete(ctx, key)
+	la.dsLk.Unlock()
 	if err != nil {
 		return fmt.Errorf("delete sampling result: %w", err)
 	}
@@ -235,5 +237,7 @@ func datastoreKeyForRoot(root *share.AxisRoots) datastore.Key {
 
 // Close flushes all queued writes to disk.
 func (la *ShareAvailability) Close(ctx context.Context) error {
+	la.dsLk.Lock()
+	defer la.dsLk.Unlock()
 	return la.ds.Flush(ctx)
 }