Apologies for being back at it again.

When testing multichannel internally, I noticed that many people instinctively looked for the channel field within the root or metadata of the AddTranscript message. Right now, channel is in both the results and metadata object. This change will mean the Python client will look for it in just the root of the message.

Accompanying change for this can be found in GitLab:
https://gitlab1.speechmatics.io/aml/aladdin/-/merge_requests/10861

Have tested the changes against the branch above, rough local testing with 5 channels transcribing the same audio file.

Without:

It
It
is up
It
It
It
is up
is up
to a new

With new changes:

channel_5: It
channel_5: is up
channel_1: It
channel_5: to a new
channel_4: It
channel_1: is up
channel_4: is up
  channel_2: It
channel_4: to a new
channel_3: It
channel_1: to a new