Update rules_jvm_external to latest (5.3)

[gregestren]

For bazelbuild/bazel#17032 (eliminate dependencies on cfg = "host")

[alexeagle]

ping @gregestren do you plan to green this up?

[gregestren]

ping @gregestren do you plan to green this up?

I was getting more dependency mismatches than I bargained for and last snapshot still had some failures.

I'll run through it again and clarify those failures. And ideally fix them if it's nothing egregious.

[gregestren]

Still iterating on this.

Updating this is much more awkward than I would have hoped:

• Focusing on failures: Android Jetpack, Android Roboelectric
• I originally tried pushing as much as I could into modules. I'm discouraged by new problems that seems to pop up, and stuff like rules_kotlin 1.9.0 notes noting the experimental nature of module support
• Trying to revert rules_android & rules_kotlin to WORKSPACE. Latest error: another dep complaining the updated version of rules_kotlin is too new. So I'll try downgrading that.
• I don't think I can easily reproduce the CI locally. I vaguely understand the CI pipeline has a bunch of custom setup, including special staging that's not reflected in the yaml configuration. I definitely don't see the same errors when trying to repro locally, e.g on my machine.

  $ cd android/jetpack-compose
  $ bazelisk build //app/src/main:app  --noincompatible_enable_android_toolchain_resolution --enable_bzlmod
  ERROR: /usr/local/google/home/gregce/bazel/examples/android/jetpack-compose/app/src/main/BUILD.bazel:22:15: While resolving toolchains for target //app/src/main:app: No matching toolchains found for types @bazel_tools//tools/android:sdk_toolchain_type.
  

• So I'm testing each diff on CI, which means I'm adding a bunch of commits and cluttering the review queue (sorry!).

Maybe this is a lesson in too much diffing at once.

[gregestren]

Green!

I'll merge momentarily unless anyone has further requests.