Move definitions of event and account_data schemas out of the C-S spec

[richvdh]

The client-server spec contains definitions of many different event types, scattered throughout the document:

• https://spec.matrix.org/unstable/client-server-api/#room-events
• https://spec.matrix.org/unstable/client-server-api/#events-1
• https://spec.matrix.org/unstable/client-server-api/#events-2
• https://spec.matrix.org/unstable/client-server-api/#events-3
• https://spec.matrix.org/unstable/client-server-api/#events-8
• https://spec.matrix.org/unstable/client-server-api/#events-9
• https://spec.matrix.org/unstable/client-server-api/#events-10
• https://spec.matrix.org/unstable/client-server-api/#events-11
• https://spec.matrix.org/unstable/client-server-api/#events-16
• https://spec.matrix.org/unstable/client-server-api/#events-17
• https://spec.matrix.org/unstable/client-server-api/#events-18
• https://spec.matrix.org/unstable/client-server-api/#events-19
• (and probably others)

I think it's confusing that we conflate the mechanism for sending and receiving events (the C-S API) with the events that can be sent. Furthermore:

• In most cases servers don't actually care about the events at all.
• These event definitions are a significant part of the length of the C-S spec, and their inclusion there makes it hard to understand and navigate the C-S spec
• It's also hard to find the definition of a given event type.
• This is less of an argument now than it used to be, thanks to MSC2844, but adding new event types shouldn't need changes to the C-S spec.

I think we could also make a similar argument for account data types (https://spec.matrix.org/unstable/client-server-api/#midentity_server, https://spec.matrix.org/unstable/client-server-api/#events-6, https://spec.matrix.org/unstable/client-server-api/#key-storage, https://spec.matrix.org/unstable/client-server-api/#events-14, etc).