Migrate multi-line-bash-related sandbox tests into runtime tests and fix multi-line issue

[xingyaoww]

What is the problem that this fixes or functionality that this introduces? Does it fix any open issues?

We are almost done with the arch refactor for Runtime (#2404). Right now we need to make sure the new runtime has feature parity with existing default ServerRuntime before we are ready to switch over to the new architecture.

This also fixes #3143

---

Give a summary of what the PR does, explaining any non-trivial design decisions

• Move all test_sandbox.py tests into test_runtime.py to test both Runtime implementation (ServerRuntime, EventStreamRuntime). It turns out there's too many things to change, it will be easier for code reviewer if i break this down into separate PRs.
• This PR migrates all the multi-line bash command tests from test_sandbox to test_runtime.
• Support multi-line bash command handling for EventStreamRuntime
• Set PS2 to empty, so there's no weird ">" after multi-line bash command execution
• Use bashlex for bash parsing to handle multi-line commands;
• Add test cases for multi-line commands, specifically reproduce the issue in [Bug]: Multi-line bash commands time out? #3143
• Add runtime tests removed in Remove remaining global config #3099.
• Use env var to change the DEBUG in logger (otherwise it is a bit cumbersome to debug)
• Merge additional bash parsing unit tests from (test) rewrite of split_bash_parsing and added unit test #3148

---

Other references

[neubig]

Looks great, thanks for the quick fix!