Refactor lane edge reference and clarify lane numbering

[j-d-b]

This PR resolves issues #66 and #67 by:

1. Removing lane_edge_reference from the lane entity
2. Adding equivalent lane_number_start_side field to the road event entity
3. Clarifying that lane_number must start at "1" for the lane closest to the roadside edge indicated by lane_number_start_side
4. Clarifying that an increase of lane_number by "1" represents moving one lane over away from the direction indicated by lane_number_start_side

[sknick-iastate]

This looks good to me. A couple of comments based on our last spec group meeting and discussing counting shoulders, bike lanes, etc in the lane number

• I'm not sure I would interpret that shoulder/non-vehicle lanes should be included in the lane number without the background discussion we have been having recently. Especially since I don't think most agencies include these in the lane numbering currently. It may also be due to us including shoulders in the lanes table. Should we add some text in the lane_number description that says this? It may also be good to say lane_id instead of just lane when referring to the numbering as well.

• Should we add a shoulder with a lane number to the examples (specifically a right shoulder to road_event_id 12345 in both the multi-point and linestring example)

[rchoyler]

Recommend NOT including shoulders in the lane numbering sequence; unless they are designed to accommodate standard vehicular travel which is equivalent to that of the main lanes (i.e. similar road surface, supports same speed limit, same legality, etc.); such as lanes designated for use only during rush hour periods, HOV/HOT lanes, etc. Perhaps "unimproved shoulders" could be designated as LS (left shoulder) & RS (right shoulder). If numeric values are required, then perhaps use 0 for the left shoulder and 99 for the right shoulder.

[j-d-b]

Note #94 completely replaces this. If #94 is not approved/merged, this PR will again be relevant. If #94 if merged, this PR can be closed.

[j-d-b]

Closed with #94 merge