Fix defm_motif_parser reading covariate names as motif terms

[gvegayon]

Problem

defm_motif_parser (used by counter_formula, i.e. defm::td_formula()) looked for motif variables with (0?)y([0-9]+)(_([0-9]+))? over the entire formula, including the x covariate (label) suffix. Any covariate name containing y<digit> changed the model silently:

Formula (order 1, Y = a, b)	Before	After
{y0} > {y0} x Day1	Motif {a+}>{a+, b+} x Day1 (wrong motif, no error)	Motif {a+}>{a+} x Day1
{y0} x Day1	Motif {a+, b+} x Day1	a x Day1 intercept
{y0} x Day5	error: column out of range	works
{y0} > {y1} x Day1	error: y1 shows more than once	works

Related parsing bugs fixed along the way:

• Dropped covariate: {y0} > {y1}x Female passed the syntax regex (\s*x\s*), but the extraction regex required \s+x\s+, so the interaction was silently ignored. The covariate name and label now come from the capture groups of the validating regex. x must be followed by whitespace ({y0} xFemale is now a syntax error rather than ambiguous). Trailing whitespace is no longer captured in the name.
• LHS accepting current time: in two-group mode, only the RHS row was validated. {y0_1} > {y1} (order 1) was accepted and silently became the intercept motif {a+, b+}. Rows >= m_order in the LHS now throw.
• Out-of-bounds write: with m_order == 1, explicit rows weren't range-checked ({y0_5} > {y1} indexed selected out of bounds). The range check now applies to both orders.

Tests

New tests/21-defm-formula-parser.cpp covers each case. It fails on master (first assertion) and passes with the fix. Full suite: 147 assertions in 23 test cases pass.

Follow-ups: a barryr header sync, then defm will require the new barry.

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 89.33%. Comparing base (8447b23) to head (5210868).

Additional details and impacted files

@@            Coverage Diff             @@
##           master      #26      +/-   ##
==========================================
+ Coverage   89.30%   89.33%   +0.03%     
==========================================
  Files          46       46              
  Lines        3170     3170              
==========================================
+ Hits         2831     2832       +1     
+ Misses        339      338       -1     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[Copilot]

Copilot review overview

🟡 Changes recommended

The checked-in standalone defm.hpp must be regenerated so its consumers receive the fix.

Review effort: Balanced
Findings: 1

Open (1)

• Regenerate defm.hpp to include the motif-search fix · New

What changed in this PR

Fixes DEFM motif parsing so covariate names cannot alter motif terms.

Changes:

• Separates motif and covariate parsing.
• Adds row validation for transitions.
• Adds regression coverage for parser edge cases.

File	Description
include/​barry/​models/​defm/​formula.hpp	Corrects parsing and row validation.
tests/​21-defm-formula-parser.cpp	Adds regression tests.
tests/​main.cpp	Registers the new tests.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[gvegayon]

Regenerated in 5210868 by running Rscript --vanilla barry-hpp.R. make barry.hpp reported the target as up to date because it has no prerequisites. defm.hpp now only scans the motif part. A standalone program using barry.hpp + defm.hpp gives 2 motif terms for {y0} > {y0} x Day1 and throws for {y0_1} > {y1}. barry.hpp also picks up the Model::sample fix from #25, which had not been regenerated either.