feat(const-fold): write prepopulated inputs, place fewer Consts + next to source not targets

[acl-cqc]

Previously mb_root_inp was used to prevent values passed to prepopulate_wire / prepopulate_inputs being written into the graph - I think at the time I did this because writing them in seemed a bit superfluous, but really it's consistent with what we do to other ops. Hence, drop the mb_root_inp.

Prior to #1474 (now merged) this broke lots of things because mb_root_inp was covering that bug. Post #1474 this breaks only a few cases in const_fold::test::test_cfg which were calling assert_fully_folded

• This was failing because this test has ops with nonlocal edges from ancestor Input nodes for which values were being provided by prepopulate_inputs. The mb_root_inp was preventing these from being written into the graph, but without that, they are written into the graph which wre are checking is fully folded and so it doesn't match the expected form. (Specifically, assert_fully_folded looks for the nodes you'd expect in a DFG; we are running it on a BasicBlock in a CFG, and the fed-in constant values are the successor-predicate)
• Hence, change the Const-insertion code. Instead of placing (duplicating) each Const next to each input using the const value, place a single Const next to the output producing the value, and route edges (perhaps nonlocal) to all targets.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 83.82%. Comparing base (4fdadd2) to head (8feb879).
⚠️ Report is 10 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1489      +/-   ##
==========================================
+ Coverage   83.81%   83.82%   +0.01%     
==========================================
  Files         187      187              
  Lines       28943    28956      +13     
  Branches    27880    27877       -3     
==========================================
+ Hits        24258    24273      +15     
+ Misses       3523     3522       -1     
+ Partials     1162     1161       -1     

Flag	Coverage Δ
python	92.33% <ø> (-0.08%)	⬇️
qis-compiler	91.66% <ø> (ø)
rust	83.49% <100.00%> (+0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.