ChannelDB.load_data: add comment re bad performance, and some speed-up

On my machine, ChannelDB.load_data() went from around 6 sec to 4 sec,
just by commenting out that assert in lnmsg.

related #6006

electrum/channel_db.py

@@ -581,6 +581,9 @@ class ChannelDB(SqlDB):
     @sql
     @profiler
     def load_data(self):
+        # Note: this method takes several seconds... mostly due to lnmsg.decode_msg being slow.
+        #       I believe lnmsg (and lightning.json) will need a rewrite anyway, so instead of tweaking
+        #       load_data() here, that should be done. see #6006
         c = self.conn.cursor()
         c.execute("""SELECT * FROM address""")
         for x in c:

electrum/lnmsg.py

@@ -57,14 +57,13 @@ def _make_handler(msg_name: str, v: dict) -> Callable[[bytes], Tuple[str, dict]]
     Returns function taking bytes
     """
     def handler(data: bytes) -> Tuple[str, dict]:
-        nonlocal msg_name, v
         ma = {}  # map of field name -> field data; after parsing msg
         pos = 0
         for fieldname in v["payload"]:
             poslenMap = v["payload"][fieldname]
             if "feature" in poslenMap and pos == len(data):
                 continue
-            assert pos == _eval_exp_with_ctx(poslenMap["position"], ma)
+            #assert pos == _eval_exp_with_ctx(poslenMap["position"], ma)  # this assert is expensive...
             length = poslenMap["length"]
             length = _eval_exp_with_ctx(length, ma)
             ma[fieldname] = data[pos:pos+length]