TransgeneExpressionHere: emit Expressed_in as [label](id) markdown

Reverts the `label----id` cell payload that v1.14.3 / v1.14.4 used
and lands on the standard `[label](id)` markdown convention
(same shape name / parent / template / dataset already use, e.g.
ListAllAvailableImages returning
  parent: '[medulla](FBbt_00003748)'
).

Click-target wiring stays out of VFBquery. The Java
VFBqueryJsonProcessor (companion v2.2.4.14 release) extracts the
`(id)` half from each clickable column's markdown and builds the
row's id column with the same `----` delimiter SOLRQueryProcessor
uses (e.g.
  upstream_class_id + delim + downstream_class_id
or
  cluster.short_form + delim + term.short_form + delim + pubs[0]
  + delim + dataset.short_form
).
Griddle then reads click targets via
  path.split(entityDelimiter)[entityIndex]
where entityIndex per visible column is declared in
queryBuilderConfiguration.js (name=0, expressed_in=1, reference=2,
dataset=3, ...).

VFBquery stays delimiter-agnostic — only lists and `[label](id)`
markdown go over the wire, same as every other clickable column
already does today.

Cypher change: the alphabetically-first leaf anat within the
Owlery closure is now emitted as
  apoc.coll.sort(collect(DISTINCT
      apoc.text.format("[%s](%s)",
                       [anat.label, anat.short_form])))[0]
                                                       AS expressed_in
and `id` stays as plain `ep.short_form` — the Java processor will
pack it together with anat.short_form (and any other per-column
ids) when it builds the row.

Verified locally against pdb on pacemaker neuron (FBbt_00006048):
  P{GAL4-per.BS}   -> '[pacemaker neuron](FBbt_00006048)'
  P{GAL4-tim.E}    -> '[pacemaker neuron](FBbt_00006048)'
  P{GSV6}GS10340   -> '[pacemaker neuron](FBbt_00006048)'
  P{cry-GAL4.E}    -> '[pacemaker neuron](FBbt_00006048)'

Patch bump 1.14.4 -> 1.14.5. Wire shape change implicitly
invalidates stale @with_solr_cache buckets carrying the v1.14.3 /
v1.14.4 `label----id` payloads.

Author: Robbie1977

setup.py

@@ -3,7 +3,7 @@
 
 here = path.abspath(path.dirname(__file__))
 
-__version__ = "1.14.4"
+__version__ = "1.14.5"
 
 # Get the long description from the README file
 with open(path.join(here, 'README.md')) as f:

src/vfbquery/vfb_queries.py

@@ -5424,15 +5424,16 @@ def get_transgene_expression_here(anatomy_short_form: str, return_dataframe=True
     INSTANCEOF leaf subclasses, not the parent. Same closure pattern as
     get_instances (v1.12.8).
 
-    v1.14.3: adds Expressed_in column matching v2 prod's single-value
-    display. Each EP row carries ONE representative leaf anatomy class
-    its Individuals are INSTANCEOF (alphabetically first within the
-    Owlery closure of the queried class), wire-formatted as
-    `label----short_form` so the geppetto-vfb QueryLinkComponent
-    (single-link customComponent that's already on this column in
-    queryBuilderConfiguration.js) parses the entity id and renders it
-    as a clickable navigation link. No new React-side infrastructure
-    or wire-format invention needed.
+    v1.14.5: adds Expressed_in column matching v2 prod's single-value
+    display. Each EP row carries one representative leaf anatomy class
+    (alphabetically first within the Owlery closure) as `[label](id)`
+    markdown — the same convention name / parent / template / dataset
+    already use. Click-target wiring for v2's QueryLinkComponent is
+    handled downstream by the Java VFBqueryJsonProcessor (which
+    extracts the `(id)` from each clickable column's markdown and
+    builds the row's id column to match SOLRQueryProcessor's pattern).
+    VFBquery stays delimiter-agnostic — only lists and `[label](id)`
+    markdown go over the wire, same as every other clickable column.
     """
     # Resolve the full subclass closure of the input anatomy class via
     # Owlery. Owlery handles OWL inference (equivalence classes, defined
@@ -5474,7 +5475,12 @@ def get_transgene_expression_here(anatomy_short_form: str, return_dataframe=True
         WHERE anat.short_form IN {anat_short_forms!r}
         WITH ep,
              collect(DISTINCT ar.pub[0]) AS pub_shorts,
-             apoc.coll.sort(collect(DISTINCT anat.label + '----' + anat.short_form))[0] AS expressed_in
+             // Alphabetically-first representative leaf anat per EP,
+             // emitted as `[label](id)` markdown — the same shape
+             // name / parent / template / dataset already use. The
+             // Java VFBqueryJsonProcessor handles click-target wiring
+             // downstream; VFBquery just emits the markdown.
+             apoc.coll.sort(collect(DISTINCT apoc.text.format("[%s](%s)", [anat.label, anat.short_form])))[0] AS expressed_in
         ORDER BY ep.label
         {limit_clause}
         CALL {{
@@ -5513,7 +5519,7 @@ def get_transgene_expression_here(anatomy_short_form: str, return_dataframe=True
     results = vc.nc.commit_list([main_query])
     df = pd.DataFrame.from_records(get_dict_cursor()(results))
     if not df.empty:
-        df = encode_markdown_links(df, ['name', 'template', 'thumbnail'])
+        df = encode_markdown_links(df, ['name', 'expressed_in', 'template', 'thumbnail'])
 
     if return_dataframe:
         return df