Fix geometry overlap: nudge dirtCentralStairwellLower3 face by 1 um

[oksuzian]

Fixes #1925.

dirtCentralStairwellLower3, added in #1918, abuts CentralStairwellLowerStairs at Mu2e z = 6375.4. That neighbour is rotated by -1.5707963268 — π/2 truncated at the 10th decimal — so its face is tilted ~5e-12 rad off square and pokes into the new volume.

bin/overlapCheck.C calls CheckOverlaps(1e-12), so the artifact is reported on the default geometry (geom_common.txt):

Overlap ov00000: HallAir/dirtCentralStairwellLower30x545f8d0
  overlapping HallAir/CentralStairwellLowerStairs0x55f22b0  ovlp=3.11047e-10

This moves the abutting face out by 1 µm so the volumes no longer touch — the same remedy already applied to the neighbour itself, whose yHalfThickness carries "remove 1 um to prevent conflict as the rotation angle is not exactly pi/2".

Validation

Regenerated mu2e_common.gdml from the patched geometry via gdmldump.fcl (art exit 0) and reran overlapCheck.C:

	illegal overlaps/extrusions
baseline (main @ 1891b328)	1
this change	0

Diffing the regenerated GDML against the baseline after stripping Geant4 pointer suffixes shows exactly the two twoDimVertex entries for this face:

-<twoDimVertex x="9956.8" y="-12369.8"/>
+<twoDimVertex x="9956.801" y="-12369.8"/>
-<twoDimVertex x="9956.8" y="-13589"/>
+<twoDimVertex x="9956.801" y="-13589"/>

Volume counts identical (9987 physvol), and the neighbour's rotation is untouched. No physics consequence — 1 µm on a dirt volume, against an original discrepancy of 3e-13 m.

Why not fix the rotation angle instead

I first tried replacing the truncated literal with a full-precision π/2, which also gives 0 overlaps. Two reasons I abandoned it:

1. SimpleConfig has no π constant and no expression evaluation — values go through strtod, so M_PI/2 isn't available. It would have to be the bare literal -1.5707963267948966, which is unpleasant to read and invites someone to "tidy" it back to something shorter later.
2. Shortening it doesn't work. I tested the intermediate -1.5707963267 (10 decimals, erring short of π/2 on the theory it would open a gap) and it still overlaps, at ovlp=5.78e-09 — worse than the original. Any inexact right angle tilts a corner in somewhere; only the full 16-digit value clears the tolerance. So the angle route is all-or-nothing on that literal.

The 1 µm nudge sidesteps the precision question entirely and matches the idiom already in these files.

Note for a follow-up

Two sibling volumes carry the same truncated literal:

• bldg/centralStairwellUpperStairs1.txt:11
• bldg/centralStairwellUpperStairs2.txt:11

Neither currently overlaps anything, so they're out of scope here. Worth knowing they're the same latent pattern if a new volume is ever placed against them. @sdifalco — these are your files, happy to defer if you'd rather handle the family differently.

[FNALbuild]

Hi @oksuzian,
You have proposed changes to files in these packages:

• Mu2eG4

which require these tests: build.

@Mu2e/fnalbuild-users, @Mu2e/write have access to CI actions on main.

⌛ The following tests have been triggered for 1cbd142: build (Build queue - API unavailable)

About FNALbuild. Code review on Mu2e/Offline.

[oksuzian]

Replaced the original approach (full-precision pi/2 on the rotation literal) with a 1 um face nudge — same result, 0 overlaps, but no 16-digit constant to maintain. Force-pushed; the PR body above describes the current change. Details on why shortening the angle literal doesn't work are in the "Why not fix the rotation angle instead" section.

[FNALbuild]

☀️ The build tests passed at 1cbd142.

Test	Result	Details
test with	✅	Command did not list any other PRs to include
merge	✅	Merged 1cbd142 at 1891b32
build (prof)	✅	Log file. Build time: 08 min 49 sec
ceSimReco	✅	Log file.
g4test_03MT	✅	Log file.
transportOnly	✅	Log file.
POT	✅	Log file.
g4study	✅	Log file.
cosmicSimReco	✅	Log file.
cosmicOffSpill	✅	Log file.
ceSteps	✅	Log file.
ceDigi	✅	Log file.
muDauSteps	✅	Log file.
ceMix	✅	Log file.
rootOverlaps	✅	Log file.
g4surfaceCheck	✅	Log file.
trigger	✅	Log file.
check_cmake	✅	Log file.
FIXME, TODO	✅	TODO (0) FIXME (0) in 0 files
clang-tidy	✅	0 errors 0 warnings
whitespace check	✅	no whitespace errors found

N.B. These results were obtained from a build of this Pull Request at 1cbd142 after being merged into the base branch at 1891b32.

For more information, please check the job page here.
Build artifacts are deleted after 5 days. If this is not desired, select Keep this build forever on the job page.

[oksuzian]

@FNALbuild run build test

[FNALbuild]

⌛ The following tests have been triggered for ea88441: build (Build queue - API unavailable)

[FNALbuild]

☀️ The build tests passed at ea88441.

Test	Result	Details
test with	✅	Command did not list any other PRs to include
merge	✅	Merged ea88441 at 1891b32
build (prof)	✅	Log file. Build time: 08 min 49 sec
ceSimReco	✅	Log file.
g4test_03MT	✅	Log file.
transportOnly	✅	Log file.
POT	✅	Log file.
g4study	✅	Log file.
cosmicSimReco	✅	Log file.
cosmicOffSpill	✅	Log file.
ceSteps	✅	Log file.
ceDigi	✅	Log file.
muDauSteps	✅	Log file.
ceMix	✅	Log file.
rootOverlaps	✅	Log file.
g4surfaceCheck	✅	Log file.
trigger	✅	Log file.
check_cmake	✅	Log file.
FIXME, TODO	✅	TODO (0) FIXME (0) in 0 files
clang-tidy	✅	0 errors 0 warnings
whitespace check	✅	no whitespace errors found

N.B. These results were obtained from a build of this Pull Request at ea88441 after being merged into the base branch at 1891b32.

For more information, please check the job page here.
Build artifacts are deleted after 5 days. If this is not desired, select Keep this build forever on the job page.