Inconsistent format for message IDs

[lgrahl]

DeliveryReceipt.message_ids contains a list of IDs in bytes whereas the various send methods of the Connection instance usually return a hex-encoded ID. This makes it really annoying to handle.

We should unify the format. My personal tendency is to use bytes. This would definitely require a major version increase because it breaks API. Any thoughts, @dbrgn?

[lgrahl]

Supplement: I believe the reason why I did this was that it's harder to read byte representations in the CLI, so maybe the CLI should still print hex-encoded IDs.

[lgrahl]

(Let's make sure to remove the deprecated DeliveryReceipt.ReceiptType.user_ack alias when releasing a new major version.)

[dbrgn]

I'd go with the type returned by the API. If the API returns hex values, we could keep them in hex, it's a bit easier to handle. But converting everyting to bytes would be fine too, as long as it' sconsistent :)

[lgrahl]

I'd go with the type returned by the API.

Well, the API just can't return bytes because HTTP. 😉

Another reason why bytes is better: Hex-encoding is ambiguous (because case doesn't matter) which is an issue when comparing two IDs.

[dbrgn]

Ok, go for bytes!