Avoid re-implementing expression simplification in pruning.rs

[xudong963]

          I think we should leave the and/or folding to a dedicated optimizer pass (IIRC that's "expression simplification") instead of re-implementing it here again. It's unnecessarily complex, potentially slow, requires more tests to write and maintain, and is a potential source of additional bugs.

Originally posted by @crepererum in #15995 (comment)

[adriangb]

I’ll point out that we also need expression simplification for filter pushdown (they need to be re cast and then re simplified against each files schema)

#15780

[xudong963]

The existing expression simplification code is used to optimize logical expr. The expr mentioned by the issue is physical expr and I think @adriangb 's expr is also physical expr.

The obvious way is to check if we can make the existing code generic