chore(mvp): cut half-baked features for mvp focus #87

As part of the discussion in #73, we need to remove some features that are currently half-baked at best in order to improve our focus on a core subset of features to let us get a shippable v1 released. This includes:

• Cutting support for on-the-fly SQL queries/streaming from the server from NATS/buffered data – only stream raw events (row insertions, etc) as they are received, no attempting to aggregate with CH data to give updated SUMs, etc. TBD on if basic WHERE filters will be allowed/applied or not, as we don't want to build a custom CH SQL parser but also understand we need some level of filtering, for both security purposes (ie restricted table or column access) and for tables with high write volumes that we cannot afford to stream raw.
• Removing the distributed/clustered approach for the time being. The plan will be to focus on the standalone ONLY, helping us speed up build times (fewer test cases and release versions) and remove complexity. If teams need more reliability than standalone (one server) can provide, even on k8s or similar, they can use our managed service at wavehouse.cloud. We will, however, still build all component integrations with highly abstracted + modularized frameworks, like we do for our dedupe layer, so that adding other DB options or extending to cluster modes in the future doesn't require a large scale refactor.
• Remove broken Github actions workflows and jobs and simplify existing ones to be more efficient + better mirror the makefile
• Remove now-deprecated config options for cluster mode and other config cleanup where applicable
• Simplify hub fanout process (wildcards) given reduced complexity for current MVP plans
• Remove stream name config options and implementation in MQ – not needed when only MQ impl. is embedded, will re-add for remote NATS when needed