Single folder containing all results

[Jammy2211]

Currently, all results are written in a structured output directory:

output
    path_0
         search_1
               ...
    path_1
         search_1
               ...
    ...

This is great, but for many datasets can be cumbersome to navigate.

We should make a second folder called output/latest_results which stores all results of all searches that are running.

The idea is that on a super computer, this folder can be downloaded and inspected, without a need to download all the results all the time.

The folder should have a results.info file, which provides a summary of each result on each line, for example:

unique_id                                     search         unique_tag                log_likeihood                iteration_numer
hifusnhtr894hnt498    search[2]_source          dataset_14                   -3973.9478                                    666

The result.info should ensure that the unique_tag is never duplicated, as for search chaining this would correspond to the same dataset in different searches being displayed.

This file should be customized via an Analysis.customize_latest_results function. This function should have a dictionary enter containing these entries:

{
"unique_id" : "hifusnhtr894hnt498",
 "search" :  "search[2]_source",
"unique_tag" : "dataset_14",               
"log_likeihood"  : "  -3973.9478",              
"iteration_numer": "666",
}

A user can then add additional entries that get displayed.

The latest_results folder should also contain .png files displaying the best-fit model so far. I don't think this requires any autofit source code development, simply that a user adds the visualization methods to Analysis.customize_latest_results.

Another benefit of the latest_results is we can make this folder even when a database is in session mode, so a user can stay in-tune with the results as they run.