Currently, the parser will check each event type definition by its name (eg. `jdk.ExecutionSample` etc.) and if an incompatible type definition is encountered, eg. coming from another concatenated chunk, a new 'synthetic' type name is created by postfixing the original type name with a unique id.

While this properly maintains the integrity of the parsed data in situation when each chunk contains events of the type in question, in certain situations when that particular event is enabled in only one the chunks (eg. when we are concatenating the built-in JFR recording and the async-profiler generated recording) this mechanism will cause unnecessary troubles as the events will not be recognized by any tool that relies on the event type name (eg. JMC UI).

In such situations, when we have events materialized for only one of the event type definitions it should be possible to run the parser in a 'relaxed' mode when the event type with no emitted events could just be ignored - thus simplifying the life of the event consumers.