Allow for 0 or 1 indexing with bu_parallel_id

Got bit by this with the new BoT raytracing - on Windows,
bu_parallel_id() starts counting from 1, so the thread indexing into the
hit array was offset accordingly.  On Linux, however, bu_parallel_id()
DOES return 0 - so that offset generated an invalid index and
segfault.

Simplest approach appears to be to increase the hit array allocation by
1, to allow either indexing approach to succeed.  Additional cost of a
single unused hit container is negligible and avoids any other need for
conditional logic.

src/librt/primitives/bot/bot.c

@@ -534,7 +534,10 @@ rt_bot_prep(struct soltab *stp, struct rt_db_internal *ip, struct rt_i *rtip)
     sps->tris = tris;
     sps->vertex_normals = tri_norms;
     sps->num_cpus = bu_avail_cpus();	// NOTE: this does NOT respect user requested cpu count (ie if -P was used)
-    sps->hit_arrays_per_cpu = (hit_da *) bu_calloc(sps->num_cpus, sizeof(hit_da), "thread-local bot hit arrays");
+
+    // We need an extra array entry here, to accommodate platforms using both 0
+    // and 1 indexing for bu_parallel_id()
+    sps->hit_arrays_per_cpu = (hit_da *) bu_calloc(sps->num_cpus+1, sizeof(hit_da), "thread-local bot hit arrays");
     bot->tie = (void*) sps;
 
     // struct bvh_build_node and struct bvh_flat_node are puns for fastf_t[6] which are the bounds
@@ -703,7 +706,7 @@ rt_bot_shot(struct soltab *stp, struct xray *rp, struct application *ap, struct
     if (UNLIKELY(!sps))
 	return 0;
 
-    int thread_ind = bu_parallel_id() - 1;  // 0-index
+    int thread_ind = bu_parallel_id();
     hit_da *hits_da = &sps->hit_arrays_per_cpu[thread_ind];
     hits_da->count = 0;
 
@@ -865,7 +868,9 @@ rt_bot_free(struct soltab *stp)
 	bu_free(sps->tris, "bot triangles");
 	bu_free(sps->vertex_normals, "bot normals");
 	if (sps->hit_arrays_per_cpu) {
-	    for (size_t i = 0; i < sps->num_cpus; i++) {
+	    // As in allocation, we have one extra entry beyond num_cpus to allow
+	    // for various platforms using either 0 or 1 bu_parallel_id indexing
+	    for (size_t i = 0; i < sps->num_cpus+1; i++) {
 		if (sps->hit_arrays_per_cpu[i].items) bu_free(sps->hit_arrays_per_cpu[i].items, "bot thread-local hit arrays");
 	    }
 	    bu_free(sps->hit_arrays_per_cpu, "bot array of dynamic thread-local hit arrays");