Extents updates

[erialC-P]

Method changes:

• Connectivity is calculated early in the workflow, and used as a mask for the entire analysis - e.g. pixels outside of the coastal mask are never used to calculate elevation or extents in the first place (vs. the original approach of running the elevation workflow on a larger set of input pixels then clipping the outputs at the end of the workflow). This is faster and more efficient - plus we can re-use the same connectivity output in the Extents analysis too
• Frequency is the only exception to the above - it's now calculated across the entire tile extent (v.s. being clipped to excluded deep ocean pixels). This makes it more useful for Extents analysis - we don't need to treat missing nodata pixels over deep water differently)
• Switched to improved ensemble tide modelling coast from dea_tools (will be replaced by eo-tides in the future)

Other:

• Updated requirements to more recent packages

[github-actions]

For full integration test results, refer to the Tests directory README.

[vnewey]

Looks good