xr.merge does not respect datatypes of inputs

What happened?

I have noticed a behaviour when using xr.merge to combine results where the datatype of the inputs are changed on the output to float.

This is undesirable at times, in particular when the results are to be used to index into an array.

What did you expect to happen?

Expected the datatypes to be preserved.

Minimal Complete Verifiable Example

subsets = []
for i in range(10):
    subsets.append(xr.DataArray([i,],dims=['idx'],coords={'idx':[i,]},name='da'))

display(subsets[0].dtype)

merged=xr.merge(subsets)

display(merged)
display(merged.da.dtype)

MVCE confirmation

• Minimal example — the example is as focused as reasonably possible to demonstrate the underlying issue in xarray.
• Complete example — the example is self-contained, including all data and the text of any traceback.
• Verifiable example — the example copy & pastes into an IPython prompt or Binder notebook, returning the result.
• New issue — a search of GitHub Issues suggests this is not a duplicate.

Relevant log output

dtype('int32')

<xarray.Dataset>
Dimensions:  (idx: 10)
Coordinates:
  * idx      (idx) int32 0 1 2 3 4 5 6 7 8 9
Data variables:
    da       (idx) float64 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0

dtype('float64')

Anything else we need to know?

No response

Environment

Start by running the minimal example with xarray's xr.merge entry point and inspect how the merged DataArray is constructed. Trace the merge path to find where the integer input is converted, then add a regression test based on the example; done means the merged data variable preserves the input integer dtype.