Optimizer rules compatibility

[czpmango]

Describe the bug (required)
This pr (#3255) should fix all related optimizer rules, because the expression related processing logic has been changed(VarPropertyExpression/InputPropertyExpression -> LabelTagPropertyExpression).

Your Environments (required)
Versions after this pr(#3255)

How To Reproduce(required)
Just check all expression related OptRule code logic.

Additional context

1. Some logic of this pr maybe also need been retested

(root@nebula) [nba]> with {a:1, b:{c:3, d:{e:5}}} as x return x.b.d.e
[ERROR (-1009)]: SemanticError: The type of `x' should be tag

2. Expression design should be reconsidered

   • reconsider data dependency design
   • avoid redundant design of expression types, eg. VarProperty/InputProperty/LabelTagProperty/LabelProperty, etc

3. Define the promises to optimizer to avoid errors like this pr

   • what types of plannode/expression should be taken into account (eg. LabelTagProperty or VarProperty in this pr)
   • data dependency and execution dependency conflict
   • column names should be set before optimization phase (this currently may cause some potential optimization rule bugs)
   • in what scenario dose reuse of executor dataset should be concerned(eg. Filter outputVar)
   • execution plans should be guaranteed to be stateless (eg. validator or planner setting some context states on which execution plans depend should be disabled)
   • expression conversion to Value, which is really different from expression rewriting, needs to be disabled or concerned. (eg. ColumnHints design)
   • clause level and planner's cutting of statements needed to be redesigned(eg. WHERE is not a clause in its own right — rather, it’s part of MATCH, OPTIONAL MATCH, START and WITH), this may improve some code logic related to DataCollect
   • ...

[Shylock-Hg]

To avoid these happen again, we could add more test cases which compare the optimized plan.