Revert result inference for scalar method

[VincentLanglet]

As recommended here: #520 (comment)

[ondrejmirtes]

Please explain in your own words why this has to be reverted.

[janedbal]

I believe it is summarized here: #520 (comment)

[ondrejmirtes]

I'm on a phone and I can't load the comment.

I'd like a brief explanation and also some tests that fail when this is not reverted.

[VincentLanglet]

Please explain in your own words why this has to be reverted.

When we're computing the template values of Query<Tkey, TValue>, TValue represent the result of the Query for the hydration mode HYDRATE_OBJECT but this hydration mode does not behave always like others.

For instance, SELECT o.bigIntColumn from ... will return

• An array of array of one string with HYDRATE_OBJECT
• An array of array of one string with HYDRATE_ARRAY
• An array of array of one int with HYDRATE_SCALAR
• An int with HYDRATE_SINGLE_SCALAR
• An array of int with HYDRATE_SCALAR_COLUM

I didn't revert HYDRATE_ARRAY, dunno if @janedbal has a counter example for this case.

I would say that this dev is working something like 90% of the time but such false-positive is non acceptable so we should keep mixed until it's fixed.

For non-regression test, we need to directly test hydration mode on different driver so I made #587 but I think @janedbal had something else in mind.

[ondrejmirtes]

Please add a test case that would fail if the changes weren't reverted.

[ondrejmirtes]

Thank you.