Is selection possible in layered facet charts based on a condition or pass op in FacetChart ?

[armsp]

I have with me a layered facet chart. The data is kinda tricky and I already use transform_filter before making the chart and choosing only one "category" (type) of data - "case" in this example.
Now what I want to do is to be able to show only those places in the facet which qualify a certain condition like max(case) > 8000wherecaseitself is time series. Since its a time series data I also need to employmaxon thevalueonly for typecasebut I am not able to figure out how to do that. I tried to usetransform` in facet but I am not quite there yet.

# Data
demo = pd.read_csv('https://raw.githubusercontent.com/armsp/covid19.in/master/datasets/clean_daily_statewise_distribution/concatenated_timeseries.csv')
demo = demo.reset_index().melt(id_vars = ['sno.', 'place',  'lon', 'lat', 'day'], value_vars=['case', 'death', 'recovery'], var_name='type', value_name='value')
# Charting
a = alt.Chart().mark_area(fill='red', fillOpacity=0.4).transform_filter(
    "datum.type == 'case'"
).transform_window(
    rolling_mean='mean(value)',
    frame=[-7, 0],
    groupby=['place']
).encode(
    x='day:T',
    y="rolling_mean:Q",
)

b = alt.Chart().mark_rule().transform_filter("datum.place == 'Tamil Nadu'").encode(
    x = 'day:T',
).transform_calculate(
    day=alt.expr.toDate('2020-05-04'),
)

alt.layer(a, b, data=demo).facet(alt.Column('place:N', title=None, sort=alt.EncodingSortField('value', op='max', order='descending')), title="Statewise Distribution",columns=4,
).configure_axis(
    grid=False,
).configure_view(strokeWidth=0).configure_title(
    fontSize=25,
    font='Courier',
    anchor='middle',
    color='gray',
    dy=-30
).resolve_scale(
    x='independent'
)

The above shows all place. When I update alt.layer... to below then I do get a subset of the facets, but the graphs are not quite right, which is something I understand, as I have not yet figured out how to do max of value for case.

alt.layer(b, d, data=demo).facet(alt.Column('place:N', title=None, sort=alt.EncodingSortField('value', op='max', order='descending')), title="Statewise Distribution",columns=4,
transform=[alt.FilterTransform({"field": "value" ,"gt": 8000})]

There is a possibility that it may be done better when data is massaged rather than the chart...but as you can imagine, the data was transformed that way to be able to easily study different types. Is there a way where I can just show the places where case is greater than 8000 or any other number?