[SPARK-46926][PS][CONNECT] Support convert_dtypes and infer_objects in fallback mode

[zhengruifeng]

What changes were proposed in this pull request?

Support convert_dtypes and infer_objects in fallback mode

Why are the changes needed?

for parity

Does this PR introduce any user-facing change?

new functions support in fallback mode, disabled by default

How was this patch tested?

added ut

Was this patch authored or co-authored using generative AI tooling?

no

[HyukjinKwon]

I think we should add them all for fallback for now without a couple of tests instead of adding each API here with each test cases because we can add the tests in the future, and ideally we should implement all API

[zhengruifeng]

sg, let me add them in another PR.

[zhengruifeng]

make a new PR
#44965

[zhengruifeng]

spark uses partial data to infer the schema, and warns

                    warnings.warn(
                        "The amount of data for return type inference might not be large enough. "
                        "Consider increasing an option `compute.shortcut_limit`."
                    )

in this case (compute.shortcut_limit < dataset size) , the inferred schema for the two new methods maybe incorrect. Do we need additional operations (e.g. fail this invocation) or just warn users in this way?

@HyukjinKwon

[HyukjinKwon]

Let's leave it with warning for now