[Tools] In tools/launch.py, correctly pass all DGL client/server env vars if udf is a multi-command

[erickim555]

Description

This PR modifies tools/launch.py so that a "multi command" udf_command is properly supported.
Specifically, this ensures that the DGL client/server env vars are:
(1) Persisted throughout all commands within the "multi command" udf_command
(2) Don't pollute the environment after udf_command is complete

Details:

# suppose udf_command is a "multi command"
udf_command = "pwd && python some/trainer.py"

# Before this change, tools/launch.py would execute the udf on each cluster host as (simplified):
cmd_to_run = "DGL_ROLE=server DGL_NUM_SAMPLER=2 pwd && python some/trainer.py"

# This is problematic, because the env-vars only apply to the first `pwd` command, and DGL breaks in
# difficult to debug ways when env vars aren't set inside some/trainer.py

# This PR fixes things so that it becomes:
cmd_to_run = "(export DGL_ROLE=server DGL_NUM_SAMPLER=2; pwd && python some/trainer.py)"

# Now, we satisfy both properties (1) and (2)

This does not break any backwards compatibility: existing single-cmd udf_command should still work exactly as before.

Addresses Issue: #3160

Testing:

Unit tests pass:

(dgl_py3) ~/code/dgl (distlaunch_chainedudf_envvars) $ PYTHONPATH=. python tests/tools/test_launch.py
......
----------------------------------------------------------------------
Ran 6 tests in 0.000s

OK

Checklist

Please feel free to remove inapplicable items for your PR.

• [x ] The PR title starts with [$CATEGORY] (such as [NN], [Model], [Doc], [Feature]])
• [x ] Changes are complete (i.e. I finished coding on this PR)
• [x ] All changes have test coverage
• [x ] Code is well-documented
• [x ] To the best of my knowledge, examples are either not affected by this change,
  or have been fixed to be compatible with this change
• [x ] Related issue is referred in this PR

Changes

[dgl-bot]

To trigger regression tests:

• @dgl-bot run [instance-type] [which tests] [compare-with-branch];
  For example: @dgl-bot run g4dn.4xlarge all dmlc/master or @dgl-bot run c5.9xlarge kernel,api dmlc/master

[VoVAllen]

Looks perfect. Appreciate it! @zheng-da Do you want to take a look?

[erickim555]

Hi @VoVAllen @zheng-da , how does this PR look? Is it good to be merged into master? Thanks!