fix: lagging nodes failed to sync (bnb-chain#1829)

FastFinality puts more infor into the header.extra field to keep vote information.
For mainnet, on epoch height, it could be 1526 bytes, which was 517 bytes before.
So the hardcoded 700 bytes for header could be no longer enough, increase it by
2 times would be enough.

this bug could cause P2P sync failure for nodes that are lagging behind, since they
would request access of ancient db, and GetBlockHeaders could be failed.

core/rawdb/accessors_chain.go

@@ -318,7 +318,7 @@ func ReadHeaderRange(db ethdb.Reader, number uint64, count uint64) []rlp.RawValu
 		return rlpHeaders
 	}
 	// read remaining from ancients
-	max := count * 700
+	max := count * 700 * 3
 	data, err := db.AncientRange(freezerHeaderTable, i+1-count, count, max)
 	if err == nil && uint64(len(data)) == count {
 		// the data is on the order [h, h+1, .., n] -- reordering needed