Investigate "Jafar" JFR parser

See also #126

https://github.com/btraceio/jafar

jfr-datasource currently uses the JMC JFR parser. The datasource is written in a very generic way to support all sorts of query, but in practice when used as a backend for https://github.com/cryostatio/cryostat-grafana-dashboard the actual queries that need to be handled is a very specific set, and certain types of data are simply not used - for example, we don't need to parse and deserialize thread objects or stack traces from the JFR data, because we don't have any visualization which will display this and therefore no query will ask for this.

The memory-mapped JFR files and low heap usage for even very large JFR files also might open the door to make #308 more feasible in practice without requiring huge memory requests for the jfr-datasource container.