BiSM_v2 publishable version -- aligned with direct-digitization BiSM_v1

Purpose
-------
BiSM_v1 constructs the completed relative-rotor scalar R_k = P_k + C_k,
digitizes R_k directly on a symmetric [-1,1) auxiliary interval, and obtains
the EQ/NEQ data correlation without inserting Malus' law.

BiSM_v2 does not derive that correlation again.  It asks what complementary
POL/COH population geometry is exposed by the same rotor description.

Event rules
-----------
The POL and COH click rules are the logical rules used in the public 2024
Sanctuary EPR program simF_b5.f.  No Malus probability, analytic cosine,
doubled harmonic, or fitted curve is used to generate the events.

Output
------
The four diagnostic populations are reported with the same common event
count N:
    EQP = N_EQP/N,   NEQP = N_NEQP/N
    EQC = N_EQC/N,   NEQC = N_NEQC/N
and
    P      = EQP - NEQP
    C_raw  = EQC - NEQC.

Only AFTER all Boolean events have been counted, the program calculates
    C_pred = -(1/4) sin(2 theta)
as an audit.  It is never used by the event-generation rules.

On the principal interval 0 <= theta <= pi, C_raw is compared directly with
C_pred.  Over the full 0 <= theta <= 2*pi traversal, the coherence coordinate
is treated as an oriented bivector coordinate:
    C_oriented =  C_raw for 0 <= theta <= pi
               = -C_raw for pi < theta <= 2*pi.
The raw EQC and NEQC populations are never changed by this orientation
convention.  The oriented coordinate is used only for the full-cycle phase
portrait and full-cycle doubled-harmonic audit.

Numerical audits
----------------
The run report records:
  * common-normalization population-sum error,
  * principal-interval C_raw versus C_pred RMS and maximum error,
  * full-cycle C_oriented versus C_pred RMS and maximum error.

Compile
-------
gfortran -O2 -std=f2008 simF_BiSM_publishable_v2.f90 -o simF_BiSM_publishable_v2.exe

Run
---
./simF_BiSM_publishable_v2.exe

Plot
----
gnuplot plot_BiSM_v2.gp
