Exceptions from write_cache_metadata() are not caught

[jc-harrison]

Describe the bug
In flowmachine.core.query.write_query_to_cache(), the call to write_cache_metadata() happens outside the try/except block, so if write_cache_metadata() raises an exception this will not be caught. As a result, neither q_state_machine.finish() nor q_state_machine.raise_error() is called, and the query state is incorrectly left as "executing".

To Reproduce

1. Run all steps in the "To reproduce" section of issue events.* tables think they're stored after resetting the cache, but Query.get_stored() disagrees #832. This will raise an error from write_cache_metadata().
2. Check the query state:

>>> dl_query.query_state
<QueryState.EXECUTING: 'executing'>

Expected behavior
The query state should be changed to "errored" if an error occurs at any point while writing the query to cache.