Unstructured scheme lazy regridding (with performance tests)

[stephenworsley]

Adds lazy regridding to the unstructured schemes equivalent to that added to main in #80. Also added the performance tests for unstructured schemes (including performance tests for lazy regridding) equivalent to those in #98 and #100.
Note that the main difference in code between this and #80 is the need to modify the iris function map_complete_blocks to be able to handle changing the number of dimensions during regridding.

[codecov]

Codecov Report

Merging #111 (9b8df45) into unstructured_scheme (13f8929) will increase coverage by 0.04%.
The diff coverage is 100.00%.

@@                   Coverage Diff                   @@
##           unstructured_scheme     #111      +/-   ##
=======================================================
+ Coverage                99.43%   99.48%   +0.04%     
=======================================================
  Files                       24       24              
  Lines                     1424     1550     +126     
=======================================================
+ Hits                      1416     1542     +126     
  Misses                       8        8              

Impacted Files	Coverage Δ
esmf_regrid/experimental/unstructured_scheme.py	100.00% <100.00%> (ø)
...nstructured_scheme/test_GridToMeshESMFRegridder.py	100.00% <100.00%> (ø)
...nstructured_scheme/test_MeshToGridESMFRegridder.py	100.00% <100.00%> (ø)
...ntal/unstructured_scheme/test__mesh_to_MeshInfo.py	100.00% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 13f8929...9b8df45. Read the comment docs.

[jamesp]

@stephenworsley looks good, a few comments inline, mostly requests for a little more documentation. Let's catch up on the phone and do a quick overview together too.

[trexfeathers]

Nice work @stephenworsley. I have quite a few comments I'm afraid! See what you think.

[trexfeathers]

Thanks for the changes @stephenworsley. Just a couple of minor comments left to address.

As already agreed: please raise an issue on GitHub and on JIRA for restructuring the tests.

We discussed offline yesterday that there was no easy solution to the confusing repetition in benchmarks/ci/esmf_regridder.py. But I very much hope we can revisit this in future, especially once we have a better solution for synthetic data generation. I'm certain that I 90% of the lines I read appear at least twice, in different places, but in slightly different orders - there must surely be a way of generalising some things here if we take a fresh look another time!

[jamesp]

Great stuff @stephenworsley, @trexfeathers!