[fix.deploy.mu.D8+] the scratch path, said twice on purpose — in code and where an operator can see it

Same defect as backend/scratch.py, at the other entry point. Measured 2026-08-11: pyscf's
cubegen writes the whole 6.7 MB cube into TMPDIR before returning it, and /tmp on this box is
a 30 GB tmpfs with a PER-USER quota. When the quota is tight the calculation dies with
`OSError: [Errno 122] Disk quota exceeded` from inside cubegen.write — an error a chemist can
do nothing with, arriving from a path that has nothing to do with them.

scratch.py already sets TMPDIR in code, so this is redundant by design: the unit file states
it so an operator reading the unit can SEE where the scratch lives instead of having to find
it in a module that a server imports for its side effect.

Worth recording because the same quota exhausted itself again today and took the shell down
for twenty minutes — the failure does not surface as "disk full" anywhere useful.

Author: ivanicu

deploy/systemd/dirac-fields.service

@@ -26,6 +26,15 @@ StartLimitBurst=5
 [Service]
 Type=simple
 WorkingDirectory=/home/ivan/dirac
+# Scratch goes to DISK, not to /tmp. Measured 2026-08-11: pyscf's cubegen writes the
+# whole 6.7 MB cube into TMPDIR before returning it, and /tmp here is a 30 GB tmpfs with
+# a PER-USER quota. With the quota tight, the same calculation dies with
+# `OSError: [Errno 122] Disk quota exceeded` from inside cubegen.write — an error a
+# chemist can do nothing with, arriving from a path that has nothing to do with them.
+# backend/scratch.py sets this in code as well, so every entry point is covered; this
+# line exists so an operator reading the unit can SEE where the scratch lives.
+Environment=TMPDIR=/home/ivan/.cache/dirac/scratch
+Environment=PYSCF_TMPDIR=/home/ivan/.cache/dirac/scratch
 ExecStart=/home/ivan/dirac/backend/env/bin/python backend/field_server.py
 Restart=on-failure
 RestartSec=5

deploy/systemd/dirac-physics.service

@@ -26,6 +26,15 @@ StartLimitBurst=5
 [Service]
 Type=simple
 WorkingDirectory=/home/ivan/dirac
+# Scratch goes to DISK, not to /tmp. Measured 2026-08-11: pyscf's cubegen writes the
+# whole 6.7 MB cube into TMPDIR before returning it, and /tmp here is a 30 GB tmpfs with
+# a PER-USER quota. With the quota tight, the same calculation dies with
+# `OSError: [Errno 122] Disk quota exceeded` from inside cubegen.write — an error a
+# chemist can do nothing with, arriving from a path that has nothing to do with them.
+# backend/scratch.py sets this in code as well, so every entry point is covered; this
+# line exists so an operator reading the unit can SEE where the scratch lives.
+Environment=TMPDIR=/home/ivan/.cache/dirac/scratch
+Environment=PYSCF_TMPDIR=/home/ivan/.cache/dirac/scratch
 ExecStart=/home/ivan/dirac/backend/env/bin/python backend/physics/server.py
 Restart=on-failure
 RestartSec=5