Do not remove binding alias in function declarations

[dragomirtitian]

Revert change that removes binding aliases from declaration files. Doing this correctly implies doing a lot more work than previously anticipated.

Fixes #56991
Fixes #56992

[jakebailey]

All of the examples that are broken feel like they should be fixable; I know some of this I've tackled in typeToTypeNode, but the same strategies probably don't work in declaration emit. I'm fine reverting this so it doesn't go out incorrectly and we can try and tackle it another day.

[jakebailey]

This also partially reverts #41044 as it removes all removal; I was actually expecting this to have caused d.ts errors because #41044 also introduced an error for unused renamings... but that didn't happen?

[dragomirtitian]

All of the examples that are broken feel like they should be fixable; I know some of this I've tackled in typeToTypeNode, but the same strategies probably don't work in declaration emit. I'm fine reverting this so it doesn't go out incorrectly and we can try and tackle it another day.

The problems are not unsolvable for sure. Even in declaration emit I think we ca solve it. But the problem is that it will introduce a lot of perf cost. Seems like it's not worth the complexity. (I belatedly arrived at the same conclusion as @andrewbranch 😅)

This also partially reverts #41044 as it removes all removal; I was actually expecting this to have caused d.ts errors because #41044 also introduced an error for unused renamings... but that didn't happen?

The error for unused renames is still there. I did not remove it. It only seems to pop up on some kinds of declarations for example function types let o: ({n : a}: any) => void;

[jakebailey]

The error for unused renames is still there. I did not remove it. It only seems to pop up on some kinds of declarations for example function types let o: ({n : a}: any) => void;

Just to point out an example from the PR, this is code that I believed would have gained an error: https://www.typescriptlang.org/play?#code/C4TwDgpgBA8lC8UDeBYAUFTUCGB+AXFAM7ABOAlgHYDmA3OllAEaGUCuAtkxKfRlgGNWnbr3QBfPulCQoAMQBMCKAAokOQiQo0o4wjACUCAHw5KIWkA

Yet, there's no error baseline. That being said, it looks like for declaration files we straight up skip checking the final list:

if (!node.isDeclarationFile) {
    checkPotentialUncheckedRenamedBindingElementsInTypes();
}

So, I think everything is fine? (Other than that I don't think it makes any sense that we still collect potentialUnusedRenamedBindingElementsInTypes in declaration files only to not read them later, which is not new.)

[jakebailey]

After checking the above and noticing #41044 (comment) saying explicitly that this does not show in dts files, I'm now confident that this PR is working correctly.