Support for != predicate in pruning predicates

[alamb]

Is your feature request related to a problem or challenge? Please describe what you are trying to do.

While evaluating queries against data stored in containers / multiple files, it is helpful to prune entire files using statistics (see #363 for more details). DataFusion has this logic already for ==, <, <=, > and >= operators in the pruning predicate.

However, as @NGA-TRAN noticed, there is no support for != at the moment.

https://github.com/apache/arrow-datafusion/blob/14f1eebef068a9e65f556ed74d2b6d98376c97f4/datafusion/src/physical_plan/parquet.rs#L683

Describe the solution you'd like
Add support and tests for != in predicate pruning logic

[jgoday]

@alamb Can I try to solve this issue ?

If I understand it correctly, for the non equal predicate the expression should be pruned if the literal value does not fall between the min and max values, Am I right ?
(I have already tried to implement it here (jgoday@3bb55a4), I can make a PR if you approve it)

[alamb]

Thanks @jgoday ! That would be great

For != I think we would prune the container (aka return false or NULL) from the predicate if the constant value did fall within the min/max bounds but I might be misunderstanding what you are saying; I'll love to check out the PR

[jgoday]

@alamb I think that I express myself incorrectly :)
Created #544 PR.