feat(api): add module emulation

[amitlissack]

Overview

In our HMG-CPX sync we discussed refactoring the hardware controller to go full on asyncio. When I've thought about how to tackle this task I've felt dread over how I'd test it. Would it be mandatory to have an OT2 and all the modules to make any progress?

So I looked into how to emulate the smoothie and modules. Happily, pyserial supports alternative underlying transport methods. Specifically over a socket.

Changelog

This PR introduces very elementary emulation of our three modules. Each module (and eventually smoothie) is a socket server to process GCODE commands. Only commands that read parameters are implemented. The emulators live in opentrons.hardware_control.emulation. An app module can be run to start all modules.

Finally, some tests are added to exercise the emulators: one for each module. A fixture starts the emulators. We create module objects using pyserial's socket url (socket://127.0.0.1:XXXX) to build the module objects.

This PR does NOT enable running the robot-server connecting to the emulators. I would like to support that eventually. I could see a docker configuration that runs the robot-server with emulators and environment set up for fast development. I would use that all the time.

Review requests

Testing: does serial comms still work?

• This is pretty hastily put together as a proof of concept.
• Do we want this? If so, should it live in opentron? I could see this being a separate package.
• I would ignore some of the hackery I added to get Thermocycler tests to run. I feel pretty strongly that the TCPoller will be modified making the hacks unnecessary.
• The new Thermocycler test fails on Windows. It's using very unix-y file descriptor techniques.

Risk assessment

Small changes to how we create Serial objects.

[codecov]

Codecov Report

❗ No coverage uploaded for pull request base (edge@3b7830d). Click here to learn what that means.
The diff coverage is n/a.

@@           Coverage Diff           @@
##             edge    #7353   +/-   ##
=======================================
  Coverage        ?   81.94%           
=======================================
  Files           ?      321           
  Lines           ?    21419           
  Branches        ?        0           
=======================================
  Hits            ?    17551           
  Misses          ?     3868           
  Partials        ?        0           

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 3b7830d...7759b0f. Read the comment docs.

[amitlissack]

Under the covers, if the url doesn't specify a protocol (ie "socket://") pyserial will Serial.

[amitlissack]

Moved here in order be able to patch in tests. Otherwise need root permission.

[amitlissack]

Add explanation for Thread.

Remove commnted out line

[sanni-t]

This is really cool!
It's a great setup. I think the things I mentioned during the review meeting about adding more gcode responses can be handled in follow-up PRs if it works for you.

[mcous]

I think this is a really good start. The idea of eventually moving this to its own repository (with a Docker container) really resonates with me, but I think this will be helpful off the bat for testing out new serial driver code

[sanni-t]

🎮

[sfoster1]

Really good start, thanks for taking this on! The code looks good to me, but we should definitely test

[sfoster1]

shouldn't these set height

[amitlissack]

It definitely should. I deliberately did not implement any command that changes the state of any of the emulators.

Looking at the c++ code made think that updating the state is not at all trivial for any of these modules and the smoothie. So left it for another day.