pan/bi: Ingest vecN directly (again)

Last time, I swear. We still generate writemasks but SSA-like ones and
do the lowering ourselves.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4288>
diff --git a/src/panfrost/bifrost/compiler.h b/src/panfrost/bifrost/compiler.h
index b8eeb64..3a895c2 100644
--- a/src/panfrost/bifrost/compiler.h
+++ b/src/panfrost/bifrost/compiler.h
@@ -53,6 +53,7 @@
         BI_CMP,
         BI_BLEND,
         BI_BITWISE,
+        BI_COMBINE,
         BI_CONVERT,
         BI_CSEL,
         BI_DISCARD,
@@ -541,6 +542,7 @@
 
 /* BIR passes */
 
+void bi_lower_combine(bi_context *ctx, bi_block *block);
 bool bi_opt_dead_code_eliminate(bi_context *ctx, bi_block *block);
 void bi_schedule(bi_context *ctx);
 void bi_register_allocate(bi_context *ctx);