Export Zonal Stat CSV for a multipolygon shp csv to an individual polygon csv

Question

I want to compute daily surface temperature over district level of india for last 5 year. but for this large feature collection it is not possible to compute it in one csv. because it sown an error as "User memory limit exceed" . So i am computing this one by one for each state and a single year...and this process took so much time...as there are 37 state an ut....SO is there any way to compute it once for all....here is my code sample.

var filter = table.filter(ee.Filter.eq("State_Name", 'ARUNACHAL PRADESH'))

var collection = ee.ImageCollection("ECMWF/ERA5_LAND/HOURLY")
                  .filterDate(startDate, endDate)
                  .filterBounds(filter)
                  .select('skin_temperature');
//Map.addLayer(collection,{},'collection')

var numberOfDays1 = endDate.difference(startDate, 'days'); 
              
var daily1 = ee.ImageCollection(
  ee.List.sequence(0, numberOfDays1.subtract(1))
    .map(function (dayOffset) {
      var start = startDate.advance(dayOffset, 'days');
      var end = start.advance(1, 'days');
      return collection
        .filterDate(start, end)
        .mean()
        .set('system:time_start', start.millis());
    })
);


function maskimg(img){
  return img.clip(filter)
  .subtract(273.15)
  .copyProperties(img, img.propertyNames());
}
var temp = ee.ImageCollection(daily1)
.filter(ee.Filter.date(startDate, endDate))
.select('skin_temperature').filterBounds(filter).map(maskimg);
//print ("skin_temperature", temp);
var bands = temp.toBands();
var zonal = bands.reduceRegions({
  collection: filter,
  reducer: ee.Reducer.mean(),
  scale: 11132,
  // tileScale: 8,
});
//Map.addLayer(temp,{},'temp')

Export.table.toDrive({
 
  description:'ARUNACHAL_PRADESH',
  collection: zonal,
  folder : 'TEMP_DAILY_2022',
  fileFormat: 'CSV',
  fileName: 'ARUNACHAL_PRADESH'
  
 });

i tried to do it one by one...changing the state name from feature collection....I just want to know that is there any way to do this process in a short way... like...looping the feature collection for each state and export csv for all state one by all.