[WIP] GPU specific kernels for MLEBABecLap

[ankithadas]

Summary

Additional background

Checklist

The proposed changes:

• fix a bug or incorrect behavior in AMReX
• add new capabilities to AMReX
• changes answers in the test suite to more than roundoff level
• are likely to significantly affect the results of downstream AMReX users
• include documentation in the code and/or rst files, if appropriate

[ankithadas]

For discussion: This PR introduces fairly substantial changes to MLEBABecLap, and I would like feedback on whether it should be split into smaller PRs. The main changes are:

1. Refactored the linop kernels into a templated form for consistency with other linops.
2. Removed the Box based kernels and replaced them with explicit index-based access (i, j, k, n).
3. Introduced a multi-array version of EBData, called EBDataArrays.
4. Added GPU-specific kernels to enable MF fusion.

[WeiqunZhang]

Yes, it's a good idea to split this into smaller PRs so that it's easier for testing correctness and performance. As for making the linop kernels into templates, unless it's necessary, I don't think we should do that.