Add 71 comprehensive tests to increase code coverage

[Delta17920]

Summary

This PR introduces a new test suite (test_new_coverage.R) containing 71 comprehensive tests targeting the changepoint regression functions. It significantly increases the package's code coverage, specifically aiming at the previously untested non-exported functions and edge cases in model selection.

Code Coverage Impact

Overall package coverage increased from 0.97% to 59.25%.

• R/CptReg.R: 0.00% ➔ 94.35%
• R/envcpt.R: 0.00% ➔ 80.45%
• src/C_cptreg.c: 1.77% ➔ 83.63%

Tests Added (71 Total)

1. cpt.reg (Input Validation & Functionality):

• Exhaustive input validation (data types, arrays, invalid penalties, methods, distributions, tol, minseglen bounds).
• Edge cases for minseglen warnings and errors.
• S4 class output verification (cpt.reg), plus tests for class=FALSE and param.estimates=FALSE.
• Accuracy checks for PELT and AMOC algorithms on AR(1) and AR(2) simulated data.
• Penalty edge cases (AIC, BIC, shape parameters for RSS and fixed variance).
• Support verification for 3D array inputs and trend data.

2. check_data & ChangepointRegression (Internal Routing):

• Matrix dimension validation and regressor vs. observation counts.
• Warning triggers for missing or duplicate intercepts.
• Method dispatching verification (AMOC vs. PELT) and cpts.only sorting logic.

3. envcpt (Model Subsets):

• Verification of model selection indexing (models = "meanar1cpt", models = c(5, 6)).
• minseglen propagation to internal calls.
• Detection accuracy within the subset models.

4. Penalty Weights & C-Level Dispatch (AICweights, BIC, CptReg_*):

• Structure, length, and summation bounds (weights sum to 1, all non-negative).
• Invalid object and matrix dimension error handling.
• Direct validation of CptReg_AMOC_Normal and CptReg_PELT_Normal.

GSoC 2026 Contributor Test Submission
Contributor: Pratik (@Delta17920)