1. The more important value is the one above those you mentioned: the overall expeditionary force ratio. The other ratios only direct the partitioning between Germany's allies, so it's not that bad. Finland was probably left out because I started with germany_1936.ai, modified it for DAIM, and based the other DAIM Germany files on this one afterwards, comparing with the standard files. I probably overlooked a few factors while doing this, but in this case it has no great effect: Germany will rarely send expeditionary forces "overseas" anyway, and by the time when there's a land connection between Germany and Finland, it's no longer an issue.
2. Germany is allowed to build those improvements earlier (1936 and 1939 AI), but I decided to forbid it for Barbarossa in order to use all resources to wear down Russia. Perhaps the extent of that limitation is too strict.
3. The numbers you've stated are right, but you've got to take into account two other things: The standard AIs use the "relative_build_scheme", DAIM Germany doesn't, and it has a noticeable effect on the percentages needed to get the right buildup. And in general, most of the troops for Barbarossa should be built before the campaign, keeping up the same pace while fighting in Russia can be dangerous for German manpower. Especially the AI is quite generous in spending manpower.
4. I lowered some of the odds strengths based on personal calculations with the average terrain modifier, and in a test game as Russia I was pleased with most of the German attacks. One general problem is that you can only define one value that the AI will use for all kind of ground units, be it mountaineers or mechanized.