Message metadata propagation with ack and nack

[ozangunalp]

Currently, message composition does not propagate message metadata.
This makes message interception (through decorators etc.) practically unfeasible.

With this change Message acks and nacks receive an additional Metadata parameter.
Existing methods for acks/nacks and composition continue to work by delegating to the new ack/nack functions.
The Message interface from the api package is still compatible with the MP Messaging Spec Message interface, verified by the MessageSpecCompatibilityTest test.

Other changes related to message ack/nack composition :

• Deprecate PublisherDecorator#decorate(String, String, boolean)
• Introduce IncomingInterceptor
• Change message interceptor method names:
  • OutgoingInterceptor#onMessage -> beforeMessageSend
  • IncomingInterceptor#afterMessageReceive
• Message Observation API: Allows a MessageObservationCollector to observe message creation, acks, nacks, and message processing duration.

Resolves #334

[codecov-commenter]

Codecov Report

Merging #2244 (6385b3d) into main (43d3c2d) will decrease coverage by 66.33%.
Report is 4 commits behind head on main.
The diff coverage is 1.81%.

❗ Current head 6385b3d differs from pull request most recent head eac70ac. Consider uploading reports for the commit eac70ac to get more accurate results

Additional details and impacted files

@@              Coverage Diff              @@
##               main    #2244       +/-   ##
=============================================
- Coverage     77.83%   11.50%   -66.33%     
+ Complexity     3700      459     -3241     
=============================================
  Files           301      301               
  Lines         12355    12416       +61     
  Branches       1580     1582        +2     
=============================================
- Hits           9617     1429     -8188     
- Misses         2018    10802     +8784     
+ Partials        720      185      -535     

Files Changed	Coverage Δ
...lipse/microprofile/reactive/messaging/Message.java	0.00% <0.00%> (-88.10%)	⬇️
.../smallrye/reactive/messaging/amqp/AmqpMessage.java	0.00% <ø> (-95.30%)	⬇️
...e/reactive/messaging/amqp/OutgoingAmqpMessage.java	0.00% <ø> (-78.27%)	⬇️
...e/reactive/messaging/gcp/pubsub/PubSubMessage.java	0.00% <ø> (ø)
...rye/reactive/messaging/jms/IncomingJmsMessage.java	0.00% <ø> (-47.37%)	⬇️
.../reactive/messaging/kafka/IncomingKafkaRecord.java	72.50% <ø> (-25.00%)	⬇️
...ve/messaging/kafka/impl/ReactiveKafkaConsumer.java	55.40% <0.00%> (-27.01%)	⬇️
.../reactive/messaging/mqtt/ReceivingMqttMessage.java	0.00% <ø> (-73.34%)	⬇️
...ye/reactive/messaging/mqtt/SendingMqttMessage.java	0.00% <ø> (-100.00%)	⬇️
...saging/providers/OutgoingInterceptorDecorator.java	0.00% <0.00%> (-100.00%)	⬇️
... and 3 more

... and 242 files with indirect coverage changes

[cescoffier]

A few words of caution:

• We need to make sure that the previous methods (where the ack is a function taking no parameter is still supported)
• We need to make sure we pass the TCK.

[ozangunalp]

This is the reason I didn't push this faster.
TCK pass, and existing Message tests as well. But I want to take some time to check the test coverage and see which cases are covered.

[cescoffier]

The commits might need to be refactored a bit.