AFDB Modelling: make regex parsing ProMod3 output more permissive

'ling' instead of 'filling' also catches 'sampling' which is used in
CloseLargeDeletions desaster recovery.

modelling/pymod/_afdb_modelling.py

@@ -609,7 +609,7 @@ def _TransferBFactors(messages, aln, model):
     # regular expression that finds stuff like: filling A.PRO59-(ESRQG)-A.ILE65
     # and directly makes stem residue numbers (59 and 65) available as groups
     cname = model.chains[0].GetName()
-    pat = f"filling {cname}\.[A-Z]+([0-9]*)-\([A-Z]+\)-{cname}\.[A-Z]+([0-9]+)"
+    pat = f"ling {cname}\.[A-Z]+([0-9]*)-\([A-Z]+\)-{cname}\.[A-Z]+([0-9]+)"
     for m in messages:
         if m.startswith("Resolved"):
             match = re.search(pat, m)