refactor(ot3_state_manager): Rework messages

[DerekMaggio]

Overview

Each message is now mapped to an int value instead using arbitrary string parsing

Related Docs

Changelog

• Remove to_message method from all Message classes
• Changed format of messages. All message are now 4 bytes long and have the following format: <message-id><message-content> where the length of message-id is a single byte and the length of message-content is 3 bytes.
• Replace to_message method and parse_message function with Message ABC.
  • Each concrete concrete class of Message is required to implement a build and to_bytes message
• The _parse_message method looks at the message-id byte and based off of the value passes message-content to the build method of the appropriate concrete implementation of Message
• Remove hashing of message value as response. Instead just return bytes value.
• Update tests accordingly

Review Requests

None

Risk

None

[sfoster1]

We have to change the way we handle the byte strings here, but this is definitely on the right track

[DerekMaggio]

@sfoster1 is the format string <H correct? Do we actually want it to be Big Endian?

int_val = 0xFF00
print(struct.pack(">H", int_val))
print(struct.pack("<H", int_val))

>>> b'\xff\x00'
>>> b'\x00\xff'

With the way we have been talking about the messages, we want the message to be formatted like the first print statement (Little Endian)

[sfoster1]

I think we're almost there but there's a couple subtle changes we should make.

[DerekMaggio]

@sfoster1 I feel like I should pull the parse method out of the Message class and make it a pure function.
I don't think that the concrete implementations of the Message class should have access to the parse function since the parse function has a larger scope than the concrete class should have.

What do you think?

[sfoster1]

This looks much much better structurally but there's some things to fix up.

[sfoster1]

Sorry - last review was supposed to be this status

[caila-marashaj]

What does Q axis represent here?

[DerekMaggio]

To my understanding, the Q axis is the high-throughput pipette axis.
https://github.com/Opentrons/opentrons/blob/edge/api/src/opentrons/hardware_control/types.py#L128

[sfoster1]

Looks great! Definitely the right way to do it, very solid