Rationalise Slits Implementations

Dodal now has three attempts at a device to control 4-blade slits sets. One from #431, one from #315 and one originally imported from Hyperion: https://github.com/DiamondLightSource/dodal/blob/main/src/dodal/devices/s4_slit_gaps.py

As a developer I would like to rationalise them into a single device so I can reduce technical debt. See additional discussion in #384. Current work for the Hypersion slits and #431 requires underlying EPICS implementations to be rationalised first. From @DominicOram: this is being tracked internally at https://jira.diamond.ac.uk/browse/I03-956. Currently unsure if the #315 slits can also operate on the same of PVs, @dan-fernandes may be able to comment.

## Acceptance Criteria
* There is one class for 4-bladed slits
* It is an ophyd-async device
* It has tests
* All beamline configs that use slits now use it