[FR]: Empty & reload enclave by watching SL script/package

[mieubrisse]

Background and motivation

I'm iterating on the quickstart Starlark script. This involves a lot of "make a change, then kurtosis run, repeat". Pains:

• My enclave name changes every time
• I have to wait the 3s for a new enclave to boot up every time I want to rerun
• Sometimes I make an immediate, dumb error (e.g. Starlark typo). So I make a half-second change, then need to wait another

Current behavior

1. Make change to Starlark script
2. Try a kurtosis run, which involves a ~3s enclave startup time before the SL even runs
3. Fix errors or add more code
4. Repeat

Desired behavior

I can tell Kurtosis to start watching a SL script or package. Any time I write a change, Kurtosis will empty out the enclave and redo everything as if I ran kurtosis run ....... I'll also need to be able to define parameters for this live-reloading, if my SL script or package is parameterized.

This will be extra useful when we have static port bindings, so that a user can set up static port bindings for an enclave + servicename + port ID, and the enclave ID won't keep changing under them.

How important is this feature or improvement to you?

Painful, the lack of this feature makes using Kurtosis frictionful.

[mieubrisse]

Related to idempotent runs (@gbouv ), and was also a request from Aptos