Docker: Restrict numpy<1.24.0 for pandas<1.1.0, h5py<3.0, mxnet

[maxhgerlach]

Checklist before submitting

• Did you read the contributor guide?
• Did you update the docs?
• Did you write any tests to validate this change?
• Did you update the CHANGELOG, if this change affects users?

Description

Fixes #3804

Review process to land

1. All tests and other checks must succeed.
2. At least one member of the technical steering committee must review and approve.
3. If any member of the technical steering committee requests changes, they must be addressed.

[github-actions]

Unit Test Results

  1 155 files  +  1 038    1 155 suites  +1 038   13h 27m 42s ⏱️ + 11h 29m 49s
     841 tests +         3       788 ✔️ +     158       53 💤  -    155  0 ❌ ±0 
23 809 runs  +21 349  17 023 ✔️ +15 329  6 786 💤 +6 020  0 ❌ ±0 

Results for commit 320e293. ± Comparison against base commit f99dfbf.

[github-actions]

Unit Test Results (with flaky tests)

  1 370 files  +  1 253    1 370 suites  +1 253   14h 42m 12s ⏱️ + 12h 44m 19s
     841 tests +         3       784 ✔️ +     154       53 💤  -    155  3 ❌ +3  1 🔥 +1 
28 317 runs  +25 857  19 817 ✔️ +18 123  8 496 💤 +7 730  3 ❌ +3  1 🔥 +1 

For more details on these failures and errors, see this check.

Results for commit 320e293. ± Comparison against base commit f99dfbf.

[maxhgerlach]

This fixes building all test images.

Going forward it will be good to find a nicer solution (i.e., update dependencies) so we will be able to test with Numpy 0.24+ in the future.

In the latest run of Build and Test heads (test-cpu-openmpi-gloo-py3_8-tfhead-keras_none-torchhead-mxnethead-pyspark3_3_1) there was a failure 3x with test_ray_elastic.py::test_fault_tolerance_hosts_added_and_removed, but I suppose that's flakiness. The previous CI pipeline run for this PR was fine there.