[Repo Assist] Implement InvCDF for Beta, F, and StudentT distributions (closes part of #262)

[github-actions]

🤖 This is an automated PR from Repo Assist.

Problem

Three continuous distributions had failwithf "InvCDF not implemented yet" stubs (and F had a commented-out partial implementation), making quantile computation completely unavailable. This is part of the broader issue #262 (Inverse CDF of distributions).

Remaining stubs addressed by this PR:

• Beta.InvCDF — no closed form in general
• F.InvCDF — no closed form; relates to the Beta distribution
• StudentT.InvCDF — no closed form; relates to the Beta distribution

Approach

Beta.InvCDF

Newton–Raphson iteration starting from the mean (α/(α+β)), clamped to (ε, 1−ε) to stay in the valid domain. Uses Beta.PDF as the derivative. Fallback to Rootfinding.Brent.tryFindRootWith if the Newton–Raphson residual is too large (handles edge cases like very small/large parameter values).

Closed-form special cases handled exactly:

• α=1, β=1 → Uniform: InvCDF(p) = p
• β=1 → InvCDF(p) = p^(1/α)
• α=1 → InvCDF(p) = 1 − (1−p)^(1/β)

F.InvCDF

Derived analytically from Beta.InvCDF using the relationship between the F and Beta CDFs:

F(d₁,d₂).CDF(x) = I_u(d₁/2, d₂/2)  where u = d₁x/(d₂ + d₁x)
→  u = Beta.InvCDF(d₁/2, d₂/2, p)
→  x = d₂ · u / (d₁ · (1 − u))
```

### StudentT.InvCDF
Derived analytically from Beta.InvCDF using the regularised incomplete beta representation of the StudentT CDF:
```
CDF(x) = 0.5 · I_h(ν/2, 1/2)  for x ≤ μ,  where h = ν/(ν + ((x−μ)/τ)²)
→  h = Beta.InvCDF(ν/2, 1/2, 2p)
→  k = −√(ν·(1−h)/h)
→  x = μ + τ·k
For p > 0.5: use symmetry about μ

No new dependencies introduced; all three implementations rely only on the existing Beta.CDF, Beta.PDF, and Rootfinding.Brent already present in the library.

Test Status

22 new regression tests added covering:

• Boundary values (p=0, p=1)
• Exact closed-form cases (Uniform, Beta(2,1), Beta(1,2))
• Symmetry checks (StudentT median = μ, StudentT InvCDF(p) = −InvCDF(1−p))
• R-verified quantiles: qt(0.975, 10) = 2.228139, qf(0.95, 3, 4) = 6.5914
• CDF(InvCDF(p)) ≈ p round-trip checks for various parameters

✅ Build: succeeded (7 pre-existing warnings, 0 errors)
✅ Tests: 1,216/1,216 passed (0 failures — all pre-existing tests still pass)

Remaining work for #262

After this PR, the remaining InvCDF stubs are:

• Gamma.InvCDF — already implemented (Newton–Raphson, present in the codebase)
• Normal.InvCDF — already implemented (Wichura algorithm)
• LogNormal.InvCDF — derived from Normal
• MultivariateNormal.InvCDF — complex, not yet addressed
• StudentizedRange.InvCDF — requires numerical integration, not yet addressed

Generated by 🌈 Repo Assist, see workflow run. Learn more.

To install this agentic workflow, run

gh aw add githubnext/agentics/workflows/repo-assist.md@e1ecf341a90b7bc2021e77c58685d7e269e20b99