[microTVM] Allow multiple runners in tuning micro models with meta-schedule

[mkatanbaf]

This PR adds support for running candidate projects on multiple physical boards when using meta-schedule to tune micro models, and helps reducing the tuning time.

[tvm-bot]

Thanks for contributing to TVM! Please refer to the contributing guidelines https://tvm.apache.org/docs/contribute/ for useful information and tips. Please request code reviews from Reviewers by @-ing them in a comment.

• cc @alanmacd, @gromero, @leandron, @mehrdadh _{See #10317 for details}

_{Generated by tvm-bot}

[mkatanbaf]

cc @mehrdadh @gromero @alanmacd @areusch

[mehrdadh]

To be able to run this test in CI with physical devices, we need to pass the serial numbers. I suggest to reuse the serial_number pytest argument. We could use this argument in the format of SERIAL_1,SERIAL_2,...,SERIAL_N when we want to tune with N number of devices and parse the string in serial_number pytest argument function and return a list of serial numbers. Then we could pass the serial numbers to this function.
Then in this function you could check if serial_number arg is an empty list and set it to ["0", "1"] if it's empty.

[mkatanbaf]

Thanks @mehrdadh, made the change to use the serial number when available. PTAL.

[mehrdadh]

Thanks for adding this support!
Looks good, we only need one change for the CI

[mkatanbaf]

@tvm-bot rerun