test(coverage): cover deadlock detector + DB_REGISTER (lock_deadlock.c 0.7%→66%, env_register.c 0%→55%)

[gburd]

Wire the two multi-process suites that reach the previously-cold deadlock
detector and process-registry crash/recovery paths into run_coverage.sh.

What

New COV_DEAD_REG block (on by default), modeled on the existing COV_REP
block (driver-per-test, per-test timeout, orphan-worker cleanup — these tests
reset TESTDIR and each spawns child tclsh via wrap.tcl, so a hung worker
must not wedge the whole run):

• Deadlock detector — dead001..dead006 spawn ddscript.tcl workers via
  wrap.tcl to build a ring/clump lock cycle; __lock_detect
  (src/lock/lock_deadlock.c) picks a victim. dead002/dead003 set
  -lock_detect so detection runs in-process in each worker (which flushes
  its own .gcda); dead001/004/005 use the standalone db_deadlock.
• DB_REGISTER — env007 + env012 use envscript.tcl to open the env with
  DB_REGISTER, "crash" a process (kill without close), then reopen with
  -recover/-failchk so a survivor detects the dead slot
  (__envreg_register/__envreg_isalive, src/env/env_register.c) and runs
  recovery.

No new test files — these tests already existed; they just were never in the
coverage run. Proc counts are trimmed to {2 4} ({4} for dead005) so each
finishes well inside the 300s timeout; the full {2 4 10} matrix adds only
minutes, no new lines.

Measured coverage (gcc --coverage, lcov from .libs)

file	before	after (line)	after (branch)
lock/lock_deadlock.c	0.7%	65.9%	43.2%
env/env_register.c	0.0%	54.7%	43.1%
env/env_failchk.c	17.4%	65.2%	38.6% (bonus, DB_REGISTER failchk path)

Tests run vs hang

All eight pass reliably under timeout (300s each), workers cleaned up between
runs:

PASS dead001  PASS dead002  PASS dead003  PASS dead004
PASS dead005  PASS dead006  PASS env007   PASS env012

At the full untrimmed {2 4 10} matrix dead003/005/006 exceed 300s (slow,
not deadlocked — the 10-proc iterations just take minutes and add no new lines),
which is why the wired proc counts are trimmed. No real bug found: the detector
resolves every cycle and picks the expected victim; DB_REGISTER correctly
detects the crashed slot and recovers.

README updated (new env var row, a descriptive block, and the least-covered
table refreshed for the three files that moved off the floor).

[gburd]

Superseded by rebased branch onto master (#72/#75); reopening.