Futility only at interim analysis

[nanxstats]

It would be useful to support designs with futility only analyses at specific interims, like gsDesign2 does.

In gsDesign2: test_lower can be a logical vector, indicating which analyses should include a futility bound.

Might be possible to do this in gsDesign by adding additional arguments similar to test_lower with careful examination.

[keaven]

Actual spec:

1. Efficacy bound required at final analyses
2. At least one of lower and upper bound required at each interim analysis.
3. Summary functions will need to be updated
4. Interim analysis numerical integration will need alteration (may be a good place to start).

[keaven]

...could be in conjunction with changing C code to Rust.

[nanxstats]

Maybe for this case, we should start with the technology we're familiar with 🤔