Wow. Unexpected. So unexpected I'm running my own 43.03 trials, with 400 female omni-legendary dwarves with full iron gear and steel pick/maple shield dueling steel pick/steel pick (maybe not the most realistic - goblins aren't likely to have legendary gear, large beasts' punches can't be parried iirc?) in 2x2 rooms. Results:
210 dwarves died over 4 days. Of the 190 surviving dwarves, 59 were using shield and pick, the other 131 being dual-pickers. That's 31%/69% in favour of dual-pickers.
Then I went back to when all the dwarves were alive, changed the weapons to copper morningstars because I could, and ran it again. That took longer and was more equal: of the 197 surviving dwarves, 71 were using star and shield.
Then used steel spears, since spears tend to get stuck. 76 dwarves out of 198 were using spear and shield.
Then since I wrote a script to change all weapons in fort, changed the spears to masterful. 72/199 for ⛭spear⛭ and shield.
Then did that for picks. 71/194 for ⛭pick⛭ and shield. Possible the no-quality pick result was unlucky? Hm. Second test resulted in 60/188 for no-quality picks. Is this iron axes and undead again?
Then, deciding I should be using something blunt because shield bashes are also blunt, did steel ⛭war hammers⛭. 87/199 for ⛭hammer⛭ and shield was surprisingly closest.
Realized I forgot to adjust sharpness with quality above, then hacked picks to ⛭axes⛭ with 10k sharpness and 4 days in 71/192 of shield/axe wielders were alive. Redid ⛭picks⛭, got 66/193, decided to call it close enough to not redo others.
Wanted to check shield and battle axe versus crossbow (steel bolts) and battle axe, at 20x2 tiles with combatants in opposite corners. To my great disappointment, no quivers in arena mode - you need to hold bolts in hand.
Finally, decided to do pick face-offs with Rocs , as their natural skills and body size might approximate a fleshy procedural (dwarves using full steel gear this time). ⛭picks⛭:
Shield and pick: 88 survived after 27 days (out of 100) (one dwarf lost both arms and tried to bite roc to death. And succeeded.)
Pick and Pick: 73 survived.
Rocs: 34 survived out of 200.
With just no-quality non-hacked picks,
Pick and shield: 85/100
Pick and Pick: 67/100.
Rocs: 45 survived out of 200.
Hm, guess there's something of an impact. You can't parry natural attacks, so this result is not surprising.
Maybe I'll do 43.04+ tests someday, but not for now. What this means, however, is a new distinction of military doctrine: Dual weapons versus weapon-users, weapon and shield versus unarmed. No science on two different weapons here, either - that might provide some truly fascinating possibilities.
For mining skill, there is gui/gm-unit (or make-legendary ) if you use dfhack. I recommend tweaking the script with few lines to apply to all units so that you wouldn't get carpet tunnel. For changing weapon, had to write my own script, as changeitem is a compiled plugin (and I don't have setup set to compile right now). Not completely certain if I did it right though can't find differences with gui/gm-editor, can provide code - got crashes when trying to arena several times in same play session after using it *shrug*