Adding Missing Associated Element/Index Types to Multiply ET Templates

[braxtoncuneo]

This pull request adds the associated types element_type and index_type to TensorMultiply and TensorMultiplyAdd.

These types are necessary for the use of operations such as TensorDivide, which access them as part of their implementation.

The absence of these associated types resulted in compilation errors when attempting to use such operations with TensorMultiply/TensorMultiplyAdd as operands (https://github.com/LLNL/RAJA/issues/1303#issue-1317442775).

Given that this PR simply adds in associated types, it is unlikely to create any unforeseen side-effects. After all, if any code previously intended to use these types, it would have failed to compile.

[braxtoncuneo]

This latest version includes element_type and index_type declarations in BinaryOperator as well.

[rchen20]

Yes, let's cut it off here, this is the last of the most critical fixes. I'll add more testing and cleanup after the release in here https://github.com/LLNL/RAJA/pull/1289.

[MrBurmark]

Does this have the camp update and warning fixes? Update: yes, let's merge and release.

[rchen20]

Does this have the camp update and warning fixes? Update: yes, let's merge and release.

It has everything in Camp up to October 7th. Is that enough?

[MrBurmark]

Ya, that should be the latest commit to main.

[rhornung67]

@rchen20 I restarted timed out Gitlab CI jobs and it is all green now. Do you want to merge?