retry-autopkgtest-regressions --blocked-by argument is confusing

[Brian Murray]

In the autopkgtest environment there were some issues with our proxy server configuration and I wanted to rerun all the ubuntu-release-upgrader tests (because it accesses changelogs.ubuntu.com) that had failed and were blocking other packages. I'd thought --blocked-by was the correct argument to use with retry-autopkgtest-regressions but it wasn't and I'm not the only person who is confused by that argument. From https://code.launchpad.net/~ubuntu-archive/ubuntu-archive-tools/+git/ubuntu-archive-tools/+merge/427241:

"Handling the "--blocked-by" confusion is another topic, that we might or might not address in a different MP:

Looking at a report like this:
https://people.canonical.com/~ubuntu-archive/proposed-migration/update_excuses_by_team.html

binutils blocking gcc-11 (11.3.0-4ubuntu1 to 11.3.0-5ubuntu1) for 4 days
Regressions
binutils/2.38.90.20220713-2ubuntu1: arm64 (log, history)

I would assume "./retry-autopkgtests --blocks gcc-11" to output some "&package=binutils&trigger=gcc-11/..." tests (this works as expected) and "./retry-autopkgtests --blocked-by binutils" (being the reverse) to output the same "&package=binutils&trigger=gcc-11/..." test, but using the reverse search key. => This is what "--blocked-by-tests binutils" is now handling properly.

IMO the "--blocked-by" parameter is confusing and should maybe be renamed. Looking at the source code, it checks for "migrate-after", so maybe it should be renamed to something like "--depends"/"--requires"/"--migrate-after"/("--after-transition"). WDYT?"