Final manuscript analyses and edits in targets pipeline

[kaitejohnson]

This PR is a catch-all for a bunch of additional analyses and changes made to the azure pipeline and targets pipeline to facilitate the final results and supplementary figures for the analysis. At a high level this includes:

• adding failure modes for fit and post process in azure pipeline (so jobs don't fail if data is missing or incomplete)
• excluding the manual replacement with hospital admissions only model only in the Hub comparison analysis (not in retrospective analysis)
• additional summary tables/supplemental figures/analyses necessary when writing the results
• closes Add hosp only to the real-time analysis in Hub comparison figure #211
• closes Get the real-time relative CRPS scores between ww and hosp model #209

Makes the 4 main figures with each of their own specific datasets for: real-time relative performance comparison (with and without ww), real-time absolute performance compared to Hub models submitted Feb-Mar 2024, retrospective relative performance comparison across 23-24 season, retrospective absolute performance compared to Hub models across 23-24 season

[kaitejohnson]

Quick question on the relative score for summary stats on improving vs worsening forecast performance, and then this is ready for review

@dylanhmorris

[dylanhmorris]

LGTM. Thanks, @kaitejohnson!

[dylanhmorris]

Noticed a couple places the calculation is right but mean_rel_{wis, crps} should be rel_mean_{wis, crps} so we don't confuse ourselves in the future.