Let get_spice_cmatrix() answer in a SPICE frame other than the instrument's #216

What

Let get_spice_cmatrix express the observation's pointing in a SPICE frame other than the instrument's, for example get_spice_cmatrix(frame='CASSINI_SC_COORD'), returning the J2000 rotation into that frame rather than into spice_frame_name.

The conversion is pxform(other_frame, spice_frame_name, t)^T . get_spice_cmatrix(time=t).

A matching set_spice_cmatrix(matrix, frame=...) would close the pair.

Why

The frame a C-kernel tabulates is usually not the instrument frame. CASSINI_SC_COORD is 119.94 degrees from CASSINI_ISS_NAC and NH_SPACECRAFT is 120.31 degrees from NH_LORRI; only Galileo SSI and the Voyager NACs coincide with the object their kernels describe. So a caller writing a corrected kernel has to convert, and a caller re-pointing an observation must not.

Both are legitimate, and which one a caller wants is a property of the call rather than of the API. Offering the frame as an argument answers the question per call: the instrument frame stays the default, because that is where a navigated offset is measured and where a round trip through the FOV has to close, and a kernel writer asks for the object frame explicitly.

Effect

Small for us -- it replaces about five lines in our C-kernel segment writer, which currently computes F = pxform(ck_frame, camera_frame, midtime) and applies F^-1 itself. The value is mostly that it makes the choice explicit at the call site instead of leaving each consumer to discover that the distinction exists.