Could the increase in attainable be attributed directly to the FEC errors ceasing, ie more bits available for 'real' data rather than error correction.
Hmmmm
Not directly. The FEC overhead is lost/wasted when there isn't any noise causing biterrors, and you don't regain the capacity for that 'real' data (which is the change that retransmission brings about - you do regain that capacity when there are no errors). It isn't regained in "actual" speed terms, and it isn't used to calculate a potential "attainable" speed either.
Indirectly, a change in the FEC framing parameters that causes a reduction in the FEC overhead can indeed improve speeds. Ordinarily, such a change would improve the "actual" speed - but if you were already at the top banded (or package) speed, then attainable would increase instead.
In your case, the "INP" and "delay" values weren't changed by DLM, so the sync should have happened with broadly similar FEC overheads ... but we don't know enough from MDWS to show it as a fact. If the overhead dropped, then (as you are at the top of a band), the attainable would rise.
On the other hand, a 10% alteration in attainable speed would be the kind of difference I'd expect from removing FEC protection altogether. It seems a bit too big to have happened with no change to the DLM parameters.
