The virtual entrypoint exposed by a smart rollup to smart contracts is not constrained, only its type

At origination time, one has to decide of a type for the virtual entrypoint smart contracts can use to send messages to a rollup.

However, the name of this virtual entry is not constrained, and in practice, a smart contract can call a smart rollup using any entrypoint they see fit (only they have to use the appropriate type). Besides, the name of the entrypoint chosen by the smart contract is not transmitted to the smart rollup.

This is not right. There are two UX possible:

1. We keep this wildcard behavior, but we transmit the name chosen by the smart contract to the smart rollup: I don’t think this is a correct UX, since (1) all these entrypoints would share the same type, and (2) we actually want to introduce other virtual entrypoints in the future.
2. We restrain the current behavior to only one entrypoint (%default?), and we forbid in the Michelson side the use of other entrypoints

I really think 2. is the correct way to go.

Assigning myself for now.