Benchmark improvements, including passing exceptions upwards

[trexfeathers]

🚀 Pull Request

This benchmark run was requested for the penultimate commit (before the What's New entry).

Clearer exceptions/errors

• ab32be1 & b793971: ensure failures - i.e. breakages rather than just regressions - are elevated out of subprocess.run, causing the workflow to fail.
• 28bb0d2: adds a step to the GHA workflow that will raise an issue when it fails, since we quite often miss this with the overnight runs (GitHub's default is to just email the last user to touch the workflow file).

CI usability improvements

• 5cd49c2 & 78b34f3: switch PR benchmarking to use asv continuous, which uses an exit code of 1 where performance regressions are found - so we're not relying on developers to manually read the GitHub workflow.
• 1ebf20e: improves how the benchmark runner prints things, specifically to make GHA easier to interpret.

Fixes

• 1684037 & 9c6809e: fixes the command for checking out the branch to benchmark - previous one was only tested on my fork.
• 57d7024: inflates some of the data sizes for benchmarks to make sure Iris still handles them as lazy data since Use real array for data of of small netCDF variables. #5229.
  • The benchmark workflow BEFORE shows the failing lazy data assertions
    (which now cause a workflow failure thanks to ab32be1 & b793971).
  • The benchmark workflow AFTER shows that the inflated data allows successful completion.
  • (An inadvertent extra inclusion was backed out in 70366a6).
• f631040: since the asv_delegated_conda.py plugin is the only reason we need specified conda_channels (Performance Shift(s): bbb53585 #5280 (comment)), I decided it was more appropriate / less misleading to hard-code the Conda channel into the plugin script itself.

---

Consult Iris pull request check list

[codecov]

Codecov Report

Patch coverage has no change and project coverage change: +0.02 🎉

Comparison is base (a49b5ac) 89.32% compared to head (21da04c) 89.35%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #5289      +/-   ##
==========================================
+ Coverage   89.32%   89.35%   +0.02%     
==========================================
  Files          89       89              
  Lines       22390    22417      +27     
  Branches     5374     5382       +8     
==========================================
+ Hits        20000    20030      +30     
+ Misses       1640     1639       -1     
+ Partials      750      748       -2     

see 10 files with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.