Fix forbidden/precision-unsafe numeric literals (5.0d-11, integer(wp), missing _wp)

[sbryngelson]

MFC convention: no d-exponent literals (5.0d-11 → 5.0e-11_wp); no precision-implicit bare literals where a real(wp) is expected (add _wp); loop counters must be plain integer, not integer(wp).

(a) d-exponent literal — src/simulation/m_bubbles_EL.fpp:1345

                    if ((1._wp - q_beta(1)%sf(i, j, k)) > 5.0d-11) then

Fix → 5.0e-11_wp. Note: the source linter's d-literal regex ([0-9]d0) only catches ...d0, so 5.0d-11 slips through. This is the only d-literal remaining in src/. Output-neutral (same value).

(b) integer(wp) loop counter — src/simulation/m_bubbles_EE.fpp:73

        integer(wp)                                            :: i, j, k, l

wp is a real kind; using it as an integer kind is wrong/non-portable. Fix → integer. Only occurrence in src/. Output-neutral.

(c) Bare literals missing _wp (add _wp, value unchanged):

• src/common/m_constants.fpp:38 — capillary_cutoff = 1.e-6 → 1.e-6_wp
• src/simulation/m_ibm.fpp:1137 — 0.4*patch_ib(...)%mass... → 0.4_wp*
• src/simulation/m_acoustic_src.fpp:351 — 1/foc_length(ai) → 1._wp/...
• src/simulation/m_ib_patches.fpp:603-604 — -0.5*length(...) / 0.5*length(...) → 0.5_wp
• src/simulation/m_ib_patches.fpp:773-774, :854-855 — 1e12 / -1e12 → 1e12_wp / -1e12_wp
• src/simulation/m_compute_levelset.fpp:151,157,241 — dist_vec(3) = 0 → 0._wp
• src/simulation/m_compute_levelset.fpp:514,555-566 — integer array literals (/1, 0, 0/) etc. assigned to real vectors → real literals

(d) HardcodedIC — golden-gated, NOT output-neutral:

• src/common/include/1dHardcodedIC.fpp:24,29 — 1 + 0.2*sin(5*x_cc(i)) etc. (bare 0.2, 5 default-real literals)
  These compile into pre_process initial conditions used by case-optimization tests. Adding _wp is value-neutral in exact arithmetic, but because these are golden-gated pre_process outputs the change may need ./mfc.sh test --generate --only <affected> if rounding shifts any digit. Flag for test regen — do not assume neutral.

Why it matters: d-literals and bare-real literals break single/mixed-precision builds (--single, --mixed) — a 1.e-6 literal is real32 regardless of wp, silently changing results in mixed mode. integer(wp) is outright type-incorrect.

---

_{Filed from a repo-wide code-cleanliness review; locations verified against master @ 40dde5e.}

Code references

• src/simulation/m_bubbles_EL.fpp:1345 — 5.0d-11 d-literal
• src/simulation/m_bubbles_EE.fpp:73 — integer(wp) loop counter
• src/common/m_constants.fpp:38 — 1.e-6 missing _wp
• src/simulation/m_ib_patches.fpp:603-604 — 0.5* missing _wp
• src/simulation/m_compute_levelset.fpp:514 — integer array literal to real
• src/common/include/1dHardcodedIC.fpp:24 — bare literals (golden-gated)