Make coordinator more responsive

- API:
	- Replace `emergencyModeStaringTime` by `emercengyModeStartingBlock`
- Synchronizer:
	- Track emergency mode starting block
- cli/node
	- Add working coordinator config
- coordinator:
	- Retry handler for synchronizer stats in case of error (instead of
	  waiting for the next block to try again)
	- On init, trigger an initial call to the handler for synced block
	  before waiting for the synchronizer, to force the coordinator to start
	  its logic even if there's no new block right after the node has been
	  started (very useful for running in testnet where the frequency of
	  blocks is variable)
	- Merge Msg for synced block and updated vars into one: `MsgSyncBlock`.

config/config.go

@@ -15,7 +15,7 @@ import (
 
 // Duration is a wrapper type that parses time duration from text.
 type Duration struct {
-	time.Duration
+	time.Duration `validate:"required"`
 }
 
 // UnmarshalText unmarshalls time duration from text.
@@ -46,7 +46,10 @@ type Coordinator struct {
 	// ProofServerPollInterval is the waiting interval between polling the
 	// ProofServer while waiting for a particular status
 	ProofServerPollInterval Duration `validate:"required"`
-	L2DB                    struct {
+	// SyncRetryInterval is the waiting interval between calls to the main
+	// handler of a synced block after an error
+	SyncRetryInterval Duration `validate:"required"`
+	L2DB              struct {
 		SafetyPeriod common.BatchNum `validate:"required"`
 		MaxTxs       uint32          `validate:"required"`
 		TTL          Duration        `validate:"required"`