Revert "derive an M matrix for IEF models"

[WesIngwersen]

I disagree with this approach now. For models with IFs we should rather derive them with a simplified version of equations 5 and 6 in the Estimating embodied environmental flows ..report

What should be done is removing the final demand term which is for scaling, or the equivalent of replacing y_d with an identity vector

The first term is the direct and indirect flows (domestic emissions only) per dollar associated with domestic total requirements.

$$\bm{M_d = B L_d}$$

The second term are the embodied flows from the use of imports by industries to make their commodities both directly (from A_m) and indirectly (by scaling it to total requirements using L_d)

$$\bm{M_{mi}} = \bm{M_m A_m L_d}$$

The third term which is in Eq 7 (not pictured above ) only applies to imports used directly by final consumers which is irrelevant to an M matrix which is rather direct and indirect flows per dollar produced.

Combining them should provide the correct M.

$$\bm{M = M_d + M_{mi}}$$

I am suggesting to revert this PR and to create a new branch to implement this new method for an M, but this is up to you to determine the best way to go about this.

[WesIngwersen]

The illogical argument before was to validate M as being between M_d and M_m. M must always be greater actually than M_d (unless there are negative total requirements or negative flows which are both unlikely), which only includes domestic emissions but ignores direct and indirect import emissions in the supply chains.

[WesIngwersen]

Adding @jvendries in case he wants to weigh in.

[bl-young]

I'll go ahead and revert this (#317) and start the PR with the new approach