Save layerwise diagnostics while running

[edoddridge]

This implements the functionality discussed in #162. The data are saved to .csv files every diagFreq seconds of model time, with one file per variable. If no value is specified for diagFreq, then no diagnostic output is produced.

The diagnostics have been somewhat incorporated into the test suite, but this integration could be tighter. Currently the min, max, and std of h, u, and v in the upper layer of the simulation are compared with blessed output stored in the repo. The means aren't compared because the mean of u and v should be ~0 in a closed domain.

If the simulation is picking up from a checkpoint (niter0!=0) then the model will attempt to append the diagnostic data to pre-existing files. This is tested by the test_gaussian_bump_continuation test. If no pre-existing file can be found a new one will be started.

If the simulation begins from niter0=0 and a pre-existing file is found, it will be overwritten, and the user will be notified through stdout.

[codecov[bot]]

Codecov Report

Merging #165 into master will decrease coverage by 0.42%. The diff coverage is 85.45%.

@@            Coverage Diff             @@
##           master     #165      +/-   ##
==========================================
- Coverage   95.23%   94.81%   -0.43%     
==========================================
  Files           5        5              
  Lines        1218     1273      +55     
  Branches       64       70       +6     
==========================================
+ Hits         1160     1207      +47     
- Misses         43       49       +6     
- Partials       15       17       +2

Impacted Files	Coverage Δ
aronnax/driver.py	93.45% <ø> (ø)	:arrow_up:
aronnax.f90	94.58% <85.45%> (-0.55%)	:arrow_down:

---

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 dd46be3...a4560bf. Read the comment docs.