[Proposal] Re-export config in v4 and synchronize main thread config and worker config

[peaBerberian]

In the RxPlayer team, we very often have to debug devices presenting some type of issue.

What we do usually is to create an RxPlayer branch (through git), push that branch to GitHub, then find a way (sometimes it's not so simple though) in the application in question to rely on the branch (we try to avoid pushing builds on GitHub, so we often rely on e.g. postinstall tricks to build the player in that application when installing / updating its dependencies).

Yet, having to dive into the application's CI / dependency system / build system is no fun and often breaks.

Diagnostics often follow similar RxPlayer code modifications, so I'm wondering if we could not bring back an idea we had in the v3 era:
The idea is to expose through an experimental export our global configuration, to allow updates of the RxPlayer config properties inside that application. Then we could add that custom logic inside the RxPlayer code and only enable it when a config property is enabled.

Config properties is not defined in the API, are comparable to globals (they don't need to be communicated at the function-level) and as such adding custom behavior based on config updates is relatively simple to do.

For example, for the now frequent need when debugging new devices of only loading segments once we now they are decipherable, we could just add in the target application (or even ask them to add temporarily):

import RxPlayerConfig from "rx-player/experimental/config";

RxPlayerConfig.update({
  LOAD_SEGMENTS_ONLY_IF_DECIPHERABLE: true,
})

Because in multithreading mode, we might have to synchronize the main thread's config to the worker's config, I also had to add some synchronization code. It is in essence the exact same problem than for our logger, so I kind of repeated the same logic here.