Run(s?) have multiple datasets associated with them

Description

"TypeError: list indices must be integers or slices, not str" is thrown when calling get_run() and get_runs(). This occurs regardless of the value of ignore_cache.

Steps/Code to Reproduce

import openml
openml.runs.get_run(run_id=1)

or

import openml
openml.runs.get_runs(run_ids=[1, 2, 3])

Expected Results

I expected to get an OpenMLRun object or collection of OpenMLRun objects corresponding to the run_id or id's that I passed to the function call.

Actual Results

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-6-e829c181e9ab> in <module>
      1 import openml
----> 2 openml.runs.get_run(run_id=1)

~/opt/anaconda3/lib/python3.8/site-packages/openml/runs/functions.py in get_run(run_id, ignore_cache)
    724     try:
    725         if not ignore_cache:
--> 726             return _get_cached_run(run_id)
    727         else:
    728             raise OpenMLCacheException(message="dummy")

~/opt/anaconda3/lib/python3.8/site-packages/openml/runs/functions.py in _get_cached_run(run_id)
    926         run_file = os.path.join(run_cache_dir, "description.xml")
    927         with io.open(run_file, encoding="utf8") as fh:
--> 928             run = _create_run_from_xml(xml=fh.read())
    929         return run
    930 

~/opt/anaconda3/lib/python3.8/site-packages/openml/runs/functions.py in _create_run_from_xml(xml, from_server)
    811 
    812     if "oml:input_data" in run:
--> 813         dataset_id = int(run["oml:input_data"]["oml:dataset"]["oml:did"])
    814     elif not from_server:
    815         dataset_id = None

TypeError: list indices must be integers or slices, not str

Versions

macOS-10.16-x86_64-i386-64bit
Python 3.8.8 (default, Apr 13 2021, 12:59:45)
[Clang 10.0.0 ]
NumPy 1.20.1
SciPy 1.6.2
Scikit-Learn 0.24.1
OpenML 0.12.2

Start in openml/runs/functions.py at get_run(), get_runs(), and _create_run_from_xml(), where the traceback shows dataset parsing failing. Reproduce with run_id=1 and run_ids=[1, 2, 3], then verify that runs with multiple associated datasets return OpenMLRun objects without the TypeError.