Are the arrows on the fixtures the direction the lamp is oriented or is that the light "throw" for lack of a better description.?
Short answer: Yes. Or No. Or yes and no.
Firstly, the direction of the arrow is generally the direction of 0° Azimuth in your photometric file (a factor that may affect this might be if you apply 'spin' to the defined luminaire, and perhaps also 'roll' - check the help system if you're thinking of doing that). You then have to understand how the fixture was orientated when tested, and whether the photometry was subsequently rotated in one or more the XYZ axes to be sure.
The "direction of the lamp" is probably a phrase that could get you into trouble trying to find a 'one size fits all' solution here, as lamps can be single or double ended, and may be vertically or horizontally orientated.
In general terms, there are two widely accepted conventions for lamp orientation in fixtures. IESNA conventions dictate that the longitudinal axis of the lamp(s) should be orientated parallel to 0° Azimuth in a file. CIE conventions dictate that the transverse axis of the lamp(s) should be orientated parallel to 0° Azimuth in a file.
In my experience, having all photometric files follow a dogmatic approach to lamp axis vs 0° Azimuth axis relationship is a sure way to create difficulties for lighting designers. To explain:
If I have a linear fluorescent wallwasher luminaire (that is, one with an asymmetric throw that I want to direct to a wall), if I use IES conventions then the main direction of the distribution will be orientated 90° away from where I 'point' the luminaire. So placing a fixture with a wall to the 'East', I have to orient the fixture to the 'North' or 'South' for it to correctly illuminate the wall.
Another example is a shoebox floodlight. Many shoebox floodlights will have two different lamp orientations inside the reflectors depending on the beam spread required. A wide 'Nema 2' or 'Nema 3' type distribution would probably be best served with the lamp in the same axis as the pole/arm/fixture direction, but a forward throw 'Nema 4' type distribution would probably be best served with the lamp transverse to the pole/arm/fixture direction. If we dogmatically follow the IESNA convention, when placing the fixtures with Nema 2 and Nema 3 beams, we'd place the fixtures and then orient them in an intuitive way (i.e. if we want the fixture to point North, we'd place our cursor North of the insertion point for correct orientation), but if we placed (or changed one of those fixtures to) a Nema 4 fixture, we'd insert the fixture and then to get it to point 'North' we'd have to place the cursor to the 'East' or 'West' of the insertion point.
Note also that when the fixture needs tilting, using the IESNA convention can often mean that you have to use AGi32's 'Roll' option instead of the more intuitive 'Tilt' option. (again, consider the linear fluorescent wallwasher option - if you had to tilt it with IESNA conventions, the tilt would actually place one end of the lamp higher than the other, which is not the result you're looking for.
I know this is a lot to take in, but it's actually fundamental to lighting design - if you don't know which way your photometry is pointing you're likely to do some great designs and find that the installed results are abysmal because the fixtures aren't placed (or can't be placed) in the way that the design laid them out.
I'd recommend looking at AGi32's help info in the 'Define Luminaire' dialog box, and also read LM:63 (it's cheap and generally pretty easy to read, especially with the diagrams that will explain the concepts you're asking about.