Validate dx/dy/dz spacing arguments across all C++ operator constructors (H2)

[gpagallo]

What type of PR is this? (check all applicable)

• Refactor
• Feature
• Bug Fix
• Optimization
• Example
• Documentation

Description

Adds input validation for the cell-spacing arguments (dx, dy, dz)
across every C++ operator constructor and AddScalarBC free function. Each
spacing is now required to be strictly positive and finite;
violations throw std::invalid_argument with a descriptive message naming
the offending parameter and value.

Previously these values were accepted unchecked. Passing dx = 0 produced
silent Inf/NaN propagation through the assembled operator: the 1-D
non-periodic Gradient and Divergence constructors both end with
*this /= dx; (gradient.cpp:298, divergence.cpp:262) and the static
periodicGrad1D / periodicDiv1D helpers used by the periodic-BC
constructors also divide by dx (gradient.cpp:91, divergence.cpp:93).
Passing a negative spacing yielded a sign-flipped operator with no warning.
This corresponds to finding H2 in the most recent codebase security
audit.

What changed

A single helper is introduced in src/cpp/utils.h / utils.cpp:

// utils.h
namespace mole {
    void check_spacing(Real h, const char* name);
}

// utils.cpp
void mole::check_spacing(Real h, const char* name) {
    if (!std::isfinite(h) || h <= 0.0) {
        throw std::invalid_argument(
            std::string("MOLE: ") + name +
            " must be a positive finite number, got " + std::to_string(h));
    }
}

The helper is then called at the top of every entry point that takes a
spacing argument — not only at leaf constructors — so a caller always sees
an error pointing at the function they actually invoked, with the correct
parameter name.

Files touched

Six source pairs are modified, covering 29 entry points:

File	Entry points validated
src/cpp/gradient.cpp (+ gradient.h)	6 ctors (1-D, 2-D, 3-D × {non-periodic, periodic}) + periodicGrad1D
src/cpp/divergence.cpp (+ divergence.h)	same 7 entry points as Gradient
src/cpp/laplacian.cpp (+ laplacian.h)	1-D, 2-D, 3-D ctors
src/cpp/robinbc.cpp (+ robinbc.h)	1-D, 2-D, 3-D ctors
src/cpp/mixedbc.cpp (+ mixedbc.h)	1-D, 2-D, 3-D ctors
src/cpp/addscalarbc.cpp (+ addscalarbc.h)	addScalarBClhs ×3, addScalarBC ×3
src/cpp/utils.cpp (+ utils.h)	New mole::check_spacing helper

The four typed Interpol classes (InterpolCtoF, InterpolCtoN,
InterpolFtoC, InterpolNtoC) and the original Interpol are unaffected —
they take no spacing argument.

Backward compatibility

• Source-compatible. No public signatures change.
• ABI-compatible. No member layout changes.
• Behavioral change. Code paths that previously accepted dx = 0,
  negative spacing, NaN or ±Inf and produced corrupted operators will now
  throw std::invalid_argument instead. This is the intended fix; affected
  callers were already producing incorrect results silently.

-->

• I have created an Issue that is paired with this PR
• Closes Operator constructors silently accept zero, negative, NaN and Inf for dx/dy/dz #387

QA Instructions, Screenshots, Recordings

Added/updated tests?

• Yes
• No, and this is why: please replace this line with details on why tests
  have not been included
• I need help with writing tests

A new GoogleTest file tests/cpp/test_spacing_validation.cpp is added with
the following cases:

Test	Assertion
SpacingValidation.GradientRejectsZero	Gradient(2, 10, 0.0) throws std::invalid_argument
SpacingValidation.GradientRejectsNegative	Gradient(2, 10, -1.0) throws
SpacingValidation.GradientRejectsNaN	Gradient(2, 10, NaN) throws
SpacingValidation.GradientRejectsInf	Gradient(2, 10, Inf) throws
SpacingValidation.DivergenceRejectsZero	analogous for Divergence
SpacingValidation.LaplacianRejectsZeroDz	3-D ctor with dz = 0 throws
SpacingValidation.RobinBCRejectsZero	1-D RobinBC with dx = 0 throws
SpacingValidation.MixedBCRejectsNegative	1-D MixedBC with dx = -0.1 throws
SpacingValidation.AddScalarBCRejectsZero	addScalarBC free function with dx = 0 throws
SpacingValidation.GradientAcceptsValidSpacing	sanity check — Gradient(2, 10, 0.1) does not throw

Tests cover each impacted entry-point family (Gradient, Divergence,
Laplacian, RobinBC, MixedBC, AddScalarBC) and each invalid value class
(zero, negative, NaN, Inf), plus a positive sanity case.

Keep-open request

• I am requesting maintainer review for keep-open.
  Reason:

Read Contributing Guide and Code of Conduct

• 📖 I have read the MOLE Contributing Guide: https://github.com/csrc-sdsu/mole/blob/main/CONTRIBUTING.md
• 📖 I have read the MOLE Code of Conduct: https://github.com/csrc-sdsu/mole/blob/main/CODE_OF_CONDUCT.md

[optional] Are there any post deployment tasks we need to perform?

None. Two minor follow-ups are worth noting but are not blocking:

1. The behavioral change (throw instead of silent Inf/NaN) is worth a
   one-line entry in the release notes for the next minor version.
2. Any downstream examples or tutorials that intentionally exercised
   pathological spacings should be updated to expect the exception.

[optional] What gif best describes this PR or how it makes you feel?

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!