[RF] copy-paste barlow-beeston to new class with explicit Conway name

[ferdymercury]

Fixes #15252

This Pull request:

Changes or fixes:

Checklist:

• tested changes locally
• updated the docs (if necessary)

[github-actions]

Test Results

0 tests   0 ✅  0s ⏱️
0 suites  0 💤
0 files    0 ❌

Results for commit c2e16eb.

[ferdymercury]

@HDembinski would you like to propose two things into this PR based on your reported issue?

• Replace current BarlowBeestonLL with exact computation
• Provide a third class using the approximation from your paper

[HDembinski]

Yes, I can do that. Would be good to have a common base class for the Conway approximation and my implementation, because they are very similar, except for the actual computation of the likelihood.

[ferdymercury]

Yes, I can do that. Would be good to have a common base class for the Conway approximation and my implementation, because they are very similar, except for the actual computation of the likelihood.

Thanks! Makes sense. Or maybe even a single class RooBarlowBeeston with a private enum member that chooses which approx to choose: 0 = Exact, 1 = Conway, 2 = Dembinski ? (And erasing RooBarlowBeestonLLConway)

Please tell us why you approve this request.