Add api docs for mutable message fields

[SimonWoolf]

...And update the descriptions of presence actions, which were not great

[splindsay-92]

In this case, I thought RT would set these to message.create when sending over the wire considering all messages before action was introduced are by definition a message.create? Or is that not the case?

[SimonWoolf]

I'm thinking here about history, realtime messages aren't an issue.

We could certainly say that every message with no action is implicitly a CREATE (and have the history api do that), but then the obvious question is, if that's the semantics we want then why do we even have an UNSET, why not make CREATE zero? @lmars you defined the enum, wdyt?

[lmars]

I think I originally added MESSAGE_UNSET with the assumption that Message.action would only be set for messages in a mutable messages namespace, but if we're always going to set it, then yes I think MESSAGE_CREATE should be the zero value.

[lawrence-forooghian]

Adding @m-hulbert to review for style

[SimonWoolf]

Adding m-hulbert to review for style

Mark has indicated he does not think that the deved team need to review all new docstrings.

I think this process is clearly broken right now. The existence of this repo is not much known to anyone but the SDK team, who have been at some point instructed to update it in lockstep with the specification repo. The reason for it being its own repo instead of a file in the specification repo seems to have been lost to history, and the apparent original idea that it would be owned and maintained by the deved team no longer seems to be tenable given the current resources available.

There is certainly still value in having a file for shared docstrings between different sdks. But it should be a file in the specifications repo, where it can be versioned along with the spec, modified by the same pr that adds the same fields to the spec, and rendered to https://sdk.ably.com for the convenience of SDK devs.

I'm going to merge this pr after some final tweaks, then do that.