Enhance ScanVertices/ScanEdges and some index selecting logics

[czpmango]

Degenerates to ScanVertices/ScanEdges only if no index exists. In other words, there will be IndexScan on the execution plan as long as there is an index available.(Fix Query1 and refactor the plan of Query2)

Query3 will prefer the node index(eg. index of player or team) over the edge index(index of serve).

// Query1
(root@nebula) [nba]> match (v:player:team) return v
[ERROR (-1005)]: Scan vertices or edges need to specify a limit number, or limit number can not push down.

// Query2
(root@nebula) [nba]> match (v:player:team) return v limit 3
+---+
| v |
+---+
+---+
Empty set (time spent 5721/6029 us)

// Query3
(root@nebula) [nba]> match (v:player:team)-[:serve]->(n) return v
+---+
| v |
+---+
+---+
Empty set (time spent 11076/11383 us)

[czpmango]

This error([ERROR (-1005)]: Scan vertices or edges need to specify a limit number, or limit number can not push down.) will only be raised when no available index and no limit clause.

[czpmango]

Closed because retesting the above behavior is expected.