Stop pitches becoming 2D sheets in tubes

[Olly Betts]

Consider a simple survey with a vertical step such as:

*data normal from to tape compass clino
0 1 20 0 0
1 2 30 - UP
2 3 20 0 0

*data passage station left right up down
0 5 5 10 0
1 5 5 10 0
2 5 5 0 10
3 5 5 0 10

This now results in a 2D sheet for tube at the vertical leg.

We used to have code which tried to fix this, but it actually just made the floor of the pitch lower and ceiling higher, which doesn't help the issue it is meant to address. To fix #73, I've just removed the bad code, but this situation out to be handled better.

The original idea was that the ceiling/floor should be extended out over/under the pitch a bit. A simple implementation of this turns the example into a "Z" shape tube though.

The special case handling also only kicks in if the leg is exactly vertical, but not one at (say) 89 degrees, which suffers from a very similar pitch flattening issue.

This is somewhat related to #74, though for a single leg short pitch you might well not start doing horizontal sections - I think some special handling for this case is still needed.