ml: Add software implementation of ML-KEM

This follows FIPS 203 relatively closely but takes some ideas from the
reference implementation.  In particular, how to avoid potential
side-channels via direct C division/modulo operations.  However, it just
uses Barrett reduction (no Montgomery reduction) and no negative
coefficients to avoid number format conversions and keep the
implementation clearer.

src/libstrongswan/Makefile.am

@@ -683,6 +683,13 @@ if MONOLITHIC
 endif
 endif
 
+if USE_ML
+  SUBDIRS += plugins/ml
+if MONOLITHIC
+  libstrongswan_la_LIBADD += plugins/ml/libstrongswan-ml.la
+endif
+endif
+
 if USE_TEST_VECTORS
   SUBDIRS += plugins/test_vectors
 if MONOLITHIC