Profile vs T-value vs Colormap

All points are here, and graphical analysis looks like this :

External points looks all OOT.

Here is the callout, so it's bilateral, and i'm pretty sure the engineer just want the big radius to be well shaped:

All points are here, and graphical analysis looks like this :

External points looks all OOT.

Here is the callout, so it's bilateral, and i'm pretty sure the engineer just want the big radius to be well shaped:

well there ya go. you have some red in there

In my opinion, that callout is ambiguous.  If it were me, I would ether bug controlling design authority for clarification, or just move forward with interpreting it at face value: exclusively interpret it to apply to the radius, which is continuous, to become a donut.  This is inherently controlled "all around", BUT clearly defined as a bilateral zone, ±0.25mm from the radius nominal.

So if it were me, I would do the following:
1: remove analysis of the base plane inside the cup.
2: prudently exclude blend radii at top of cup.
3: measure it as individual radii, then construct a scan set of all the radii combined, bestfit align to the combined radii set, then produce one profile of a surface to that set.

that sounds exhausting. I don't think he's wrong scanning the entire surface as hs has it set up now. i'd def increase the scan density so I can see what's going on though

less exhausting than over-inspecting and rejection of potentially acceptable goods?

but why measure as individual radii? 

why not measure the flat/plane, level and origin, inspect profile/radii as shown with scans, and report the profile? what does best fitting to the radii do?

Paste with pattern FTW! 
If there are datums on the part perhaps be aligned to them, I get trying to best fit to the radius itself but that is rarely what the enginerds intend when they call it out like that. They would typically say they want it comparted to the nominal model location, which when you machine the part the datums establish the nominal model location in reality. Checking it to a datum will also allow you to give the machinist meaningful feedback about their part thus allowing them to make corrections. 

Definitely steer WAY clear of the start/stop of the radius for your pickups though, that will throw things off. 

there arent datums on the part. EDIT: referenced on the drawing provided. 

well, since the flat/plane is not a defined datum, nor part of the feature of size being controlled... it's adding noise to the interpretation of the part and the software's ability to bestfit the profile measurement/interpretation as communicated on the drawing. 

If you had a width measurement: would you level and datum to a potentially non-perpendicular plane, before measuring that width? why not?  (It's rhetorical, you wouldn't because it can introduce error in your resultant values if the (incorrectly applied) datum plane skews your distance.)

yes but the plane can be a known value since you can measure the thickness. using the plane as a stable feature to base the rest of the profile off of isn't an unreasonable method of evaluating the rest of the profile. 

I agree with louisd , based on that drawing, I wouldn't  have included the plane for the profile measurement because the leader line only points to the radius and does not include the all around symbol. 

Either way (whether you include the plane or not), you need to do a full 3D best-fit alignment using the vector min/max fit on all of your measured points before reporting T values or creating the point colormap.  Only then would the results be equivalent to a legacy form only profile (again with the vector min/max fit) or a geometric tolerance profile command.