Allow hints for upcasting parquet to arrow integer types

[gruuya]

Which issue does this PR close?

Closes #6891.

Rationale for this change

Allow casting integers so long as the precision isn't lost.

What changes are included in this PR?

Add possible matches for up-casting all integer DataTypes.

Are there any user-facing changes?

None.

[tustvold]

I'm pretty lukewarm on baking in schema coercion into the parquet reader at this level, the interpretation of how to map the parquet data types to arrow is gnarly enough without needing to also perform non-trivial schema adaption. I think it would be better to handle this at a higher level, e.g. as proposed by #6735

[gruuya]

baking in schema coercion into the parquet reader at this level

Oh I think I understand. My initial impression was that ArrowReaderOptions::with_schema is the intended mechanism for overriding the Parquet level schema

arrow-rs/parquet/src/arrow/arrow_reader/mod.rs

Lines 276 to 283 in 54dccad

	/// Provide a schema to use when reading the parquet file. If provided it
	/// takes precedence over the schema inferred from the file or the schema defined
	/// in the file's metadata. If the schema is not compatible with the file's
	/// schema an error will be returned when constructing the builder.
	///
	/// This option is only required if you want to cast columns to a different type.
	/// For example, if you wanted to cast from an Int64 in the Parquet file to a Timestamp
	/// in the Arrow schema.

So the nuance is that in the original issue the goal was to supply a hint to a Parquet with missing type hints, whereas in my case I want to override an already-hinted Parquet schema, thus somewhat abusing/overloading this mechanism? This would also suggest that the above description needs to be changed as it is misleading.

I guess then that #6735 could also provide a solution to this problem (albeit with more effort). Let me take a look at it.

[alamb]

There is a related discussion on:

• Introduce ProjectionMask To Allow Nested Projection Pushdown datafusion#2581