Parker weighting should not throw away the first and the last projections

[SimonRit]

In RTK, we currently assume that the acquisition arc started / stopped at the first / last projection. If you refer to Parker's article, $\beta=0$ and $\beta=2\delta-\beta$ for the first and the last projections and $w=0$ in equation (12). We could instead start / stop the acquisition half an angular step before / after. This would require revisiting the angular weighting of the first / last projection.