Further to Bob's observations ..
at least equal to indicates a line in the sand ... if conditions are better, that's a bonus. I suggest that you are reading too much into the words.
if the aircraft's landing climb performance on one engine matches or exceed the missed approach climb gradient This is a black hole problem which I gave up arguing with DCA et al years ago. The general CASA Standards philosophy is that there is enough data in the OEM POH for the pilot to adjust the weight to achieve whatever gradient might be the go on the day. If you believe that, I have a bridge to sell you ...
OEI in a light twin, unless the aerodrome is very much terrain benign, presumes that, once you commit to the approach with a low actual .. then you are committing to land if it turns worse than you expected. A better risk management decision probably is to run for cover elsewhere ...
As an anecdote, on my initial issue Class One (today's MECIR) the DCA Examiner, who was a very practical and risk conscious chap, put me into an ILS at a terrain critical aerodrome and, not far from the minimum, announced that the cloud base was now fluctuating at and below the minimum (we had had the obligatory engine failure just prior to the exchange) ...
"What are your intentions ?"
Simple answer "continue to the minimum and then land" ..
"But what if you are not visual .. what will you do then ?".
Simple answer "continue to and below the minimum and land"
Inevitably, the call at the minimum was "nil sighting".
He eventually let me out from under the hood at about 50-60 ft above the runway and we duly landed. Now, he was aware of my instrument background (I had been a then unusual airline intake chap in that I didn't have a rating and had, at the time of the present test, about four years on F27 and L188 on a Second Class rating) and certification performance expertise. In any event, passed OK so it was a nil problem situation.
If you put yourself intentionally in a light aircraft OEI approach situation, with any likelihood of needing to go to the miss, then your flight management is lacking sorely (unless the bird is on fire and, then, it just isn't your day).
As to 2.5% OEI on the typical twin, do have a look in the OEM POH, if it has much useful data at all .. and check out just how much weight you would have to be under gross to get away with it ....
As to arbitrarily raising the minimum on the fly ... on what basis might you propose to do this ? First, you don't know what obstacles are critical in most cases and, in any case, you WON'T have the data required to construct the likely flightpath to determine a reasonable height increment to keep you nice.
For both the above situations, a significant problem is that you have no idea what the reconfiguration will take. Another anecdote .. years ago, I used to fly a very nice Shrike Commander. Coming back into EN one afternoon, I thought to check this situation .. with the OK for a missed approach, AND THIS WAS AEO, I commenced the miss from over the threshold with climb power only ... took the entire runway at low level to get reconfigured and to a suitable climb away speed. Now, just have a think about what the situation MIGHT be like OEI ... an absolute world of hurt.
What saves us for almost all occasions .. is the fact that we are probably going to be AEO.
For the exam, Bob's answer is the only reasonable expectation that CASA could anticipate, I suggest.