Add PushFilterIntoTableScanRule

[MingyuGuan]

• table.json example:

  {
  "name": "Parts",
  "type": "custom",
  "url": "src/main/resources/test_table.txt",
  "factory": "husky.sql.calcite.table.SimpleTestTableFactory",
  "columns":[
    {
        "id": "1",
        "name": "TRANS_ID",
        "datatype": "int"
    }, {
        "id": "1",
        "name": "ITEM_ID",
        "datatype": "int"
    }, {
        "id": "2",
        "name": "ITEM_COUNT",
        "datatype": "int"
    }, {
        "id": "3",
        "name": "PRICE",
        "datatype": "float"
    }
  ]
  }

• SimpleTestTable.java:

1. Column class to store column info;
2. Read table.json by SimpleTestTable(String tableJsonPath);
3. Read table url for data to setRows(String tableUrl);
4. setProtoRowType();

Add PushProjectIntoTableScanRule.

• Inside HuskyLogicalTableScan, we maintain a "fields" member to record project fields.

• PushProjectIntoTableScanRule matches by operand(HuskyLogicalCalc.class, operand(HuskyLogicalTableScan.class, none()).

• If HuskyLogicalCalc is trivial (the transformed program merely returns its input and doesn't exist filter), we drop calc and transform to HuskyLogicalTableScan; else, transform to new calc.

(NEW)Add PushFilterIntoTableScanRule.

• Inside HuskyLogicalTableScan, we maintain a "condition" member to record filtering condition.

• The "condition" should be conjunctive normal form, e.g. "(a AND b) OR c" will be converted to "(a OR c) AND (b OR c)".

• To computer cost of HuskyTableScan, need to consider factors of selectedFields(Project) and selectivity of rows(Filter).

[zzxx-husky]

Files under calcite/target are the outputs of compiler, which are NOT necessary to be put into the PR?

[MingyuGuan]

Files under calcite/target are the outputs of compiler, which are NOT necessary to be put into the PR?

Removed. Thanks.