Qiskit-terra: Usage of `hpc` parameters clarification

[diego-plan9]

What is the current behavior?

It seems that the hpc parameters received by transpiler.compile(..., hpc=..., ...) are not really added to the resulting Qobj - they are checked inside the function, but not used after that. This behaviour is also present in 0.5.7, and it seems we lost that functionality at some point in the past (with no strong side effects it seems :) )

Now that some changes to the transpiler are being done in #705 and related efforts, could we try to clarify if those parameters are really meant to be in use anymore, and act accordingly - either adding them to the returned object, or removing the checks from the code? There is also code in ibmqjob.py that parses them, that might need to be checked depending on the action taken.

Steps to reproduce the problem

What is the expected behavior?

Pinging explicitly @ajavadia and @jaygambetta for it!

Suggested solutions

[jaygambetta]

I would like to remove all hpc and that if it is needed it goes into a config.py as getting designed by @chriseclectic for qiskit aer

[ajavadia]

Added to #704's Transpiler epic.
This is legacy and we should remove it. I think the hpc backend should do the optimizations by default. If there is really a use case for users turning off those parallelization optimizations, then it should be done via the config as Jay mentioned.

[delapuente]

This affects #657, it should be safe to remove passing the hpc parameters.

[ajavadia]

Fixed via #724.

[diego-plan9]

Reopening as it seems that #724 does not solve the issue and hpc is still present in a number of places of qiskit/transpiler/_transpiler.py - Ali, can you check if you were referring to other issue instead?

[jaygambetta]

I think this should be closed and tracked in #1052 @diego-plan9 do you agree

[diego-plan9]

Yeah, sounds fine!