Add tofloat(Point)

[Sbozzolo]

It is easy to go from a number to a Geometry.Point, but it is not easy to go in the other direction (unless the type of point of point is known). This PR adds new methods to Float32, Float64, and BigFloat to X/Y/Z/Lat/Long point to convert them into numbers.

This feature would be very useful in converting a Field of ZPoints into a scalar field, which is what we need for the interpolation routines

[dennisYatunin]

Overloading the constructors for primitive types is a bit risky. Is there a downside to defining a function that does this, like Geometry.first_coordinate(::Point)?

[Sbozzolo]

Overloading the constructors for primitive types is a bit risky. Is there a downside to defining a function that does this, like Geometry.first_coordinate(::Point)?

This function also converts to the specified float type. This is not strictly needed, but it is nice to have. Geometry.coordinate already exists but doesn't return a float, it returns a Point, so maybe I could define tofloat

[dennisYatunin]

Overloading the constructors for primitive types is a bit risky. Is there a downside to defining a function that does this, like Geometry.first_coordinate(::Point)?

This function also converts to the specified float type. This is not strictly needed, but it is nice to have. Geometry.coordinate already exists but doesn't return a float, it returns a Point, so maybe I could define tofloat

Yes, a more "standard" way to implement this would be something like Geometry.tofloat(::Type{T}, ::Point), with a default value for T based on the Point type. Adding constructors to the float types implies that our points are "equivalent" to floats at a low level, which is not quite the case.

[Sbozzolo]

Overloading the constructors for primitive types is a bit risky. Is there a downside to defining a function that does this, like Geometry.first_coordinate(::Point)?

This function also converts to the specified float type. This is not strictly needed, but it is nice to have. Geometry.coordinate already exists but doesn't return a float, it returns a Point, so maybe I could define tofloat

Yes, a more "standard" way to implement this would be something like Geometry.tofloat(::Type{T}, ::Point), with a default value for T based on the Point type. Adding constructors to the float types implies that our points are "equivalent" to floats at a low level, which is not quite the case.

I added a function tofloat. It returns the same type as the Point, which is fine. If we need more, we can add it.

[charleskawczynski]

With the two changes I suggested, I'm fine with this PR.