Add 2d FFT adjoints

[WardBrian]

Summary

This is a continuation of #2750. The reverse mode specializations introduced in that PR for the 1D FFT generalize in the obvious way to 2 dimensions.

For FFT2: adjoint(x) += size(y) * ifft2(adjoint(y)), iFFT2: adjoint(y) += (1 / size(x)) * fft2(adjoint(x))

Tests

Existing tests in test/unit/math/mix/fun/fft_test pass

Side Effects

None.

Release notes

Added reverse-mode specializations for fft2 and inv_fft2

Checklist

• Math issue (inverse) FFT analytic adjoint-Jacobian  #2740

• Copyright holder: Simons Foundation

  The copyright holder is typically you or your assignee, such as a university or company. By submitting this pull request, the copyright holder is agreeing to the license the submitted work under the following licenses:
  - Code: BSD 3-clause (https://opensource.org/licenses/BSD-3-Clause)
  - Documentation: CC-BY 4.0 (https://creativecommons.org/licenses/by/4.0/)

• the basic tests are passing

  • unit tests pass (to run, use: ./runTests.py test/unit)
  • header checks pass, (make test-headers)
  • dependencies checks pass, (make test-math-dependencies)
  • docs build, (make doxygen)
  • code passes the built in C++ standards checks (make cpplint)

• the code is written in idiomatic C++ and changes are documented in the doxygen

• the new changes are tested

[WardBrian]

This PR speeds up gradient evaluation time on the HoloML model by about 20% (from 0.53 seconds per eval to 0.43 seconds, as reported at the start of sampling)

[bob-carpenter]

looks great. thanks!

[bob-carpenter]

this is ok to merge when tests pass