TREC Clinical Document Parsing Logic Bug

**Describe the bug**
When parsing TREC Clinical XML documents the following [line](https://github.com/allenai/ir_datasets/blob/c13285be41fc0fadeb6a2c7fdb77ff3c350d3b14/ir_datasets/datasets/clinicaltrials.py#L84C8-L84C71) is used:
```
condition = xml.find('.//condition')
condition = ''.join(condition.itertext()) if condition else ''
```
it seems that even when `condition` is a valid node it still evaluates to `False` so that condition is always set to "". Just as a test I added this line:
```
assert bool(condition) == condition is not None
```
after `xml.find` and it fails. The fix is just:
```
condition = ''.join(condition.itertext()) if condition is not None else ''
```
I can raise a PR, but wasn't sure how the tests should be changed.

**Affected dataset(s)**
TREC Clinical 2021 & 2022

**To Reproduce**
Steps to reproduce the behavior:
1. Try parsing a TREC Clinical document with the code from [here](https://github.com/allenai/ir_datasets/blob/c13285be41fc0fadeb6a2c7fdb77ff3c350d3b14/ir_datasets/datasets/clinicaltrials.py#L76).
2. Even if a valid condition tag is present the condition value will be set to "".

**Expected behavior**
The condition tag should be properly read and included in the document.

**Additional context**
Also worth mentioning some documents have multiple condition tags which you might want to include.