fix: h3 datagrams must not be sent until both sides agree on it

[JPDye]

https://www.rfc-editor.org/rfc/rfc9297#section-2.1.1:
QUIC DATAGRAM frames MUST NOT be sent until the SETTINGS_H3_DATAGRAM setting has been both sent and received with a value of 1.

Changed the SharedState so that enable_datagram represents the negotiated SETTINGS_H3_DATAGRAM (an AND of ours and the remotes).

Also fixed some broken duvet comments and added duvet tracking of h3-datagram and the associated RFC.

[vcarus]

Two of the RFC 9297 §2.1.1 lines this PR adds to the duvet snapshot ("The value of the SETTINGS_H3_DATAGRAM setting MUST be either 0 or 1" and "... MUST terminate the connection with error H3_SETTINGS_ERROR") are cheap to implement while you are here: Settings::decode in h3/src/proto/frame.rs never looks at a setting's value, and SettingsError::InvalidSettingValue(SettingId, u64) is already defined with a Display impl but is never constructed anywhere in the crate. A check for SettingId::H3_DATAGRAM with a value other than 0 or 1 returning that variant would let both citations be marked as implemented instead of registered.