Fix docs about using timespan fields in expressions.

"datetime_begin" and "datetime_end" are old and no supported in all
contexts, while "timestamp" is just wrong, and dimension-qualification
being optional is only sometimes true.

doc/lsst.daf.butler/queries.rst

@@ -425,11 +425,11 @@ Few examples of valid expressions using some of the constructs:
 
     (visit = 100 OR visit = 101) AND exposure % 2 = 1
 
-    visit.datetime_begin > T'2020-03-30 12:20:33'
+    visit.timespan.begin > T'2020-03-30 12:20:33'
 
-    exposure.datetime_begin > T'58938.515'
+    exposure.timespan.begin > T'58938.515'
 
-    visit.datetime_end < T'mjd/58938.515/tai'
+    visit.timespan.end < T'mjd/58938.515/tai'
 
     ingest_date < T'2020-11-06 21:10:00'
 

python/lsst/daf/butler/cli/opt/options.py

@@ -277,9 +277,11 @@ def _config_split(*args: Any) -> dict[str, str]:
     "--order-by",
     help=unwrap(
         """One or more comma-separated names used to order records. Names can be dimension names,
-                metadata names optionally prefixed by a dimension name and dot, or
-                timestamp_begin/timestamp_end (with optional dimension name). To reverse ordering for a name
-                prefix it with a minus sign."""
+        metadata field names, or "timespan.begin" / "timespan.end" for temporal dimensions.
+        In some cases the dimension for a metadata field or timespan bound can be inferred, but usually
+        qualifying these with "<dimension>.<field>" is necessary.
+        To reverse ordering for a name, prefix it with a minus sign.
+        """
     ),
     multiple=True,
     callback=split_commas,