Wish list: Add item to refit only nodes which have data

[garland-culbreth]

Changes

Add a new wish list item RE only fitting nodes which have data, and use node's prior predictions as its fit if there is no data for that node.

It's possible for a node which has no data to have its fit differ significantly from its prior despite having no new data, which this would prevent.

Checks

bin/run_xrst.sh passed with the following output:

$ bin/run_xrst.sh
xrst --local_toc --target html --html_theme sphinx_rtd_theme --index_page_name at_cascade 
Using following input_files: git ls-files
sphinx-build -b html -j 1 build/rst build/html
rm -r build/html/_sources
cp -r build/rst/_sources build/html/_sources
xrst: OK
run_xrst.sh: OK
$ 

[bradbell]

This is closely related to the prior predictions (because they are done by doing a fit ignoring the data). In other words, if we can predict for a region no has no data without doing a no data fit, then we can use the same technique to do priior predictions without a no data fit.

I think this can all be done by chosing the proper ancestor for the predictions. We should test that this gives the same result as the current prior predict.

Note that if we take the approach above, the fit_node_name would not be the same as the node we are predicting for, but one might have been able to do a fit with no data,

[bradbell]

I think that this item has been completed; see 08-18 on https://at-cascade.readthedocs.io/2024.html#mm-dd-08-18

Please test this and if you agree, close this pull request.

[garland-culbreth]

Pulled down the changes and bin/check_all.sh passed OK on my device, all good!