It can get a bit confusing .. some things to consider (and I don't propose to hold myself out as having any particular knowledge of JARs as opposed to FARs)
(a) if the AFM data is factored, then the 50/150 is in the grid and the wind is entered as observed. Simple way to tell if the factor is in is to see if the wind grid (carpet) lines are discontinuous (ie have a kink) at nil wind .. if the grid is unfactored, then the curves will be smooth as you go through the nil wind case. In the latter case, you would need to apply the factor prior to using the chart.
(b) I can't ever recall seeing anything definitive re approach margins (not to say such doesn't exist) and I suggest that the thing is more a risk management approach to the problem .. considerations -
(i) excess speed additives .. main problem is over run on a critical length runway .. if you run some simple calcs for the 1.67 case, and typical Vref, the Vref+20 max idea seems to make reasonably good sense.
(ii) one should always strive NOT to be under Vref at all .. the margins aren't all that great to start with and, on some aircraft, a few knots under can make a big difference to the pilot pucker factor .. 727-200 comes to mind
(c) other things which may need to be looked at .. pitching moment problems (probably only a concern for takeoff in excessive tailwind on susceptible Types), missed approach pilot control loads, (if the AFM only gives you 10 kt tail and you propose to increase that) some requirements in Part 25 are based on 10 kt and may require a revisit if one intends to extend beyond 10 kt.
Two concerns ..
(a) .. but I do have the JAR-OPS, which says that an airplane should be able to stop within 60% of the LDA(Jets)...
If you believe that (uncritically) then I have a bridge to sell you. The normal heavy certification basis is that the demonstrated LD is factored by 1.67 to arrive at the LDR. Keep in mind that
(i) the flight test program is geared to get the "best" data within the bounds of test pilot and program integrity .. let me put it this way .. the most terrifying experiences I have ever had in an aeroplane have been performance landing tests with a TP motivated by a desire to get "good" data .. There will be far more trials done than required .. the TP is quite at liberty to nominate a few to be discarded (no prizes for guessing which ones get their marching orders). Then the aerodynamicists throw out a few more (likewise).
(ii) I know that the design rules say something along the lines of "normal pilot capabilities" ... mmmm
(iii) take it as near gospel .. if you are on a limiting runway and approach through the AFM gate, on speed, you have a near guaranteed likelihood of seeing the dirt at the other end .. intimately.
That is to say, be very careful using, eg, unfactored QRH data without thinking carefully what you are doing ..
(b) ...So operational you should put a half headwind component or 150% of the tail wind to you Vref. ex. Vref: 142kts headwind 12kts Vapp: 148kts...
This is worth discussing further ... somewhere between misguided and dangerous, I would suggest ? More to the point, in my view, it is irrelevant to the approach additive question and only applies to the distance calculations ... ?
I'm being very serious in hoping that this thread develops as there is a number of aspects of approach and landing number crunching which are not all that well understood by the fraternity ...