TBD: Support dynamic experiments via REST calls

[x-ji]

Previously, the Phoenix channel was used for both dynamic experiments and interactive experiments. An experiment would only start when the designated number of participants is reached.

Now, we have a <variant-chain-generation> tuple for dynamic experiments, and an interactive experiment would only start whenever the number of players is reached.

We can simplify the mechanism at first. If there is no need for a dynamic experiment to use socket features, let's use the socket exclusively for interactive experiments.

AC:

• Dynamic experiments are no longer facilitated via channels, but rather with simple REST API calls to assign the tuple to the participant.

[x-ji]

Not sure if we're actually going to do this. TBD.