automatic handling of concurrent connections limit

I'm getting errors when I try to open 48 IMERG time steps through opendap. I don't think this is a bug, but it's not clear how I can configure xarray to avoid it. The server doesn't want to be overwhelmed by multiple data layer requests, which is reasonable, but I'm not asking for big data here. It would be fast enough if we downloaded the data one time step at a time.

urls = [build_IMERG_url(half_hour, run, version) for half_hour in half_hours]
imerg = xr.open_mfdataset(urls) # Reaches the concurrent connection limit
precipitation = imerg['precipitationCal']
daily = precipitation.mean('time') * 24

Can xarray avoid this error automatically? If not, is there a way to limit my creation of concurrent connections?

I don't know if using the netcdf library directly avoids this issue. I did try to write a for loop to connect to opendap 1 at a time with xr.open_dataset(), but if still gave the concurrent connections error. Obviously, I could just download all the data files instead of using opendap to subset, but this is not the desired outcome.

Example URL:
https://gpm1.gesdisc.eosdis.nasa.gov:443/opendap/GPM_L3/GPM_3IMERGHHL.06/2020/260/3B-HHR-L.MS.MRG.3IMERG.20200916-S000000-E002959.0000.V06B.HDF5

Start by reproducing the concurrent-connection error with xr.open_mfdataset() on the listed IMERG URLs, then compare it with the xr.open_dataset() loop described in the issue. Trace the open_mfdataset entry point and its remote dataset handling. Done means users can limit or avoid concurrent OpenDAP connections without downloading all files, with coverage for the reported case.