corrected profile_index to 1 indexed

[callumrollo]

Fixes Issue #40

I noticed this as processing datasets in chunks of N dives would always produce N-1 profiles. profile_index is given a value of 0 when inflecting (see discussion in #40) but the first profile is given a value of 0. Later when profiles and grids are extracted form the timeseries, this first valid dive is treated as an inflection point and ignored.

This PR:

• starts dives from 1 rather than 0 (heresy for Python I know, but necessary in this case)
• corrects the comment of profile_index to correctly note that samples during inflection have profile index 0, not N + 0.5
• removed the function get_profiles which is not used anywhere in the codebase

This should fail the tests for profile_index in slocum and seaexplorer