MML-294 Fixing some issued with the UIAction and Dialog elements

[symphony-mariacristina]

Goal of this PR is to fix some minor issues found during validation.

• Allowing the presence of whitespaces inside the Dialog element
• If a UIAction with a target-id referring to a Dialog comes in the messageML before the Dialog itself, the message should be allowed and parsed correctly.
• No more that one form is allowed inside a Dialog element
• We should be able to match Dialog-UIAction target id even if they are nested inside another element.
• Matching Dialog-UiAction should be in the same scope (at the same level of the messageML tree) to be valid.

Closes #294

[symphony-mariacristina]

I'm wondering if this post processing is a bit too heavy for each message. A possible improvement could be to add a flag to the messageML to be set when a UIAction with target-id is found during messageML parsing. In this case we only do this post processing when we know for sure that there's a match to be found. We already do something similar for chimes https://github.com/symphonyoss/messageml-utils/blob/master/src/main/java/org/symphonyoss/symphony/messageml/elements/MessageML.java#L130 Any thoughts @symphony-youri @symphony-thibault ?

[symphony-youri]

I'm wondering if this post processing is a bit too heavy for each message. A possible improvement could be to add a flag to the messageML to be set when a UIAction with target-id is found during messageML parsing. In this case we only do this post processing when we know for sure that there's a match to be found. We already do something similar for chimes https://github.com/symphonyoss/messageml-utils/blob/master/src/main/java/org/symphonyoss/symphony/messageml/elements/MessageML.java#L130 Any thoughts @symphony-youri @symphony-thibault ?

This is just about navigating a few elements in the tree right? I would say usually messages don't have many elements (like hundreds max?) so it should be pretty fast. But if you want to confirm that you can use JMH to assess the impact (with/without post processing) (MessageMLContextBenchmark)

[symphony-mariacristina]

I did a test using the JMH with a complex message and seems like the post processing is not impacting the performance of the MessageML parsing, so I guess we can keep it as it is.

[symphony-mariacristina]

do you get different results now?

Not so much fortunately :)

Results with the post processing:

Benchmark                                                     Mode  Cnt    Score    Error  Units
MessageMLContextBenchmark.parseComplexMessageMLWithEntities  thrpt   25  826.608 ± 19.860  ops/s

Results without:

Benchmark                                                     Mode  Cnt    Score    Error  Units
MessageMLContextBenchmark.parseComplexMessageMLWithEntities  thrpt   25  822.082 ± 20.188  ops/s