Allow FieldNames to override the field name of ContextJsonProvider when used in the context of IAccessEvent

[brenuart]

Closes #658

[brenuart]

@philsttr Looks like the field thread could also be promoted to a common name. The ThreadNameJsonProvider could then be used for both logging and access events.

Similarly, SequenceJsonProvider and UuidProvider are currently limited to ILoggingEvent but may potentially be useful in the context of IAccessEvent as well.

[philsttr]

@philsttr Looks like the field thread could also be promoted to a common name. The ThreadNameJsonProvider could then be used for both logging and access events.

The *FieldNames is only needed/used for the providers that are in LogstashAccessEncoder and LogstashEncoder by default. *FieldNames is needed because the default providers of those encoders are not configured individually (like is done for the *CompositeJsonEncoder). Since ThreadNameJsonProvider is not in LogstashAccessEncoder by default, it doesn't need to be in *FieldNames.

For comparison, when the *CompositeJsonEncoders are used directly, *FieldNames is not used. Instead, the field name used by each provider is configured directly on each provider, as opposed to being configured on the encoder level.

Similarly, SequenceJsonProvider and UuidProvider are currently limited to ILoggingEvent but may potentially be useful in the context of IAccessEvent as well.

Yes, they can be refactored to handle IAccessEvent or ILoggingEvent, but they still wouldn't need to to be in *FieldNames, since they would not be in LogstashAccessEncoder or LogstashEncoder by default.

[brenuart]

Ok - got it for Field Names.

@philsttr Should we open separate issues to refactor ThreadNameJsonProvider, SequenceJsonProvider and UuidProvider to be usable for both ILoggingEvent and IAccessEvents ?
I'm not convinced many people will use them with IAccessEvents (should anybody use access events at all...) but this could make the code base consistent. It wasn't clear to me at first sight why these providers were limited to logging events... In fact, there is no reason except nobody asked for it ;-)

[philsttr]

Yes, please open an issue for them. I agree we should make them consistent. Thanks :)