Generalize kill-zone/tracer limits; size streamfunction zones adaptively

[hdrake]

Summary

The streamfunction flux arrays and their per-zone loops were hard-coded to a fixed 21 kill zones, and the geographical kill-zone (10) and tracer (10) capacities were scattered as bare literals across many files (including the per-project kill_zones.F90). This generalizes those limits into named constants and sizes the streamfunction allocation to the zones a run actually uses.

This is the first of two planned changes; a follow-up will reduce the online streamfunction memory footprint (the 0:ntracmax flux dimension) using the maxlbas machinery introduced here.

Changes

• Named capacity constants in mod_precdef: MAXGEOZONES, MAXTRACERS, and MAXZONES = 1 + MAXTRACERS + MAXGEOZONES, used to dimension the kill-zone and tracer configuration arrays.
• Adaptive maxlbas: the actual number of streamfunction zones is computed from the run — namelist geographic zones + subdomain wall zones + tracer kill zones — and finalized in init_subdomain once exitType and all zones are known. The offline flux arrays and the compute_stream loop now size to maxlbas instead of a fixed 21.
• Derive from SIZE(): runtime loop bounds and the subdomain "last 4" wall slots are derived from SIZE() of the arrays, removing hard-coded slot numbers (7–10) and 1,10 loops in mod_subdomain, kill_zones.F90 (×5), mod_diffusion, and mod_init.
• Well-defined defaults: ienw/iene/jens/jenn are initialized to 0 so "undefined slot" is well-defined.

Bug fixes

• Latent out-of-bounds when the configured zone count differs from 21.
• Under-allocation when an exitType=2 run uses a subdomain (which promotes exitType to 3 and adds geographic walls) — now correct because maxlbas is finalized in init_subdomain after the promotion.

Behavior

Unchanged for valid configs. The streamfunction output (psi_*.csv) now contains the actual number of zone blocks instead of always 21 (e.g. 2 for a single kill-zone, 9–11 for subdomains). Verified byte-identical to the previous output for the real (non-empty) zones; the dropped trailing zones were always zero.

Testing

• Clean build (Theoretical project, gfortran -fbounds-check), no warnings.
• Regression: single kill-zone → maxlbas=2, psi_* byte-identical to baseline for real zones.
• Subdomain box → maxlbas=11; full-width subdomain (E/W walls skipped) → maxlbas=9; both run clean under bounds-checking.
• The change is size-preserving by construction (MAXGEOZONES=MAXTRACERS=10, MAXZONES=21).
• Note: the tracer-array changes compile clean and are size-preserving, but were not exercised by a tracer-enabled run (the Theoretical project ships no tracer fields).

🤖 Generated with Claude Code

---

Addresses #19.

[hdrake]

Indirectly addresses #5 by making the dimensions of the output more intuitive.