fix(external): fetch and build each clone once per build pass

`crafter-build test` builds a project's tests side by side, and a
project whose tests each copy its externalDependencies hands every one
of them the same unpinned dependency. Each test's Build() pulled the
shared clone. Before the clone had a lock, those pulls ran at the same
time and broke its .git/FETCH_HEAD, which failed every test of
Crafter.Graphics (issue #34). Since 653dd0b they wait for each other,
but every one is still a network round trip: 26 tests cost 26 pulls of
each dependency, one after another, and for a CMake dependency as many
configure checks and builds.

depResults already records one build pass, keyed by the PcmDir of every
dependency configuration built in it. An external dependency now goes
in the same map, keyed by the directory its build leaves
(ExternalBuildDir: the clone, or for CMake its build directory). The
first configuration that needs it fetches and builds it; the others
wait for that and take its outcome. Keying on the build directory
rather than the clone keeps a CMake declaration from taking the outcome
of a header-only declaration of the same clone, which only fetched it.
A fresh depResults is a fresh pass, so the next build still pulls, and
every caller of Build() gets this: RunTests, a plain build, and the
pragma dependencies LoadProject builds.

To make that possible BuildExternal is split. BuildExternal is now the
shared work, lock, fetch, configure and build, and returns the error.
ExternalOutputs computes what one declaration compiles and links with,
because two declarations of one clone can differ in includeDirs,
libDirs and libs; ExternalBuildResult loses its error field. The work
no longer stops between steps when the configuration that started it
is cancelled: others may be waiting on it, and a run cut short would
leave them a clone nobody fetched. A dependency configuration's Build()
already runs to completion for the same reason.

The promise and future the dependency threads used to build or wait for
a configuration are now one helper, OncePerPass, which both kinds use.
An exception thrown in BuildExternal used to escape its thread into
std::terminate; it is now a build error, as for a dependency.

With eight tests that share a local dependency, a test run fetched it
eight times before this and fetches it once now (SharedExternalFetch,
next commit).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

AGENTS.md

@@ -86,7 +86,7 @@ Conventions the existing tests follow:
 - Collect failures with a `Check(cond, msg)` helper that prints `FAIL: ...` to stderr and counts, then return nonzero at the end. Do not stop at the first failure unless later steps are meaningless without it.
 - Drive the engine in-process: `Build(cfg, depResults, depMutex)`, `RunTests`, the lint driver. Do not shell out to `bin/crafter-build` and grep its stdout.
 - Fixtures under `tests/<Name>/fixture/` are inputs. A test that edits or touches files first stages a copy under `fs::temp_directory_path() / "crafter-build-<name>"` and removes it on the next run. Variant sources are checked in as `*.in` files and copied over the original inside the staged copy.
-- Use a fresh `depResults` map per build pass: the map memoises each configuration's result for one pass, so reusing it skips rebuilds you meant to observe. `Build()` itself is safe to call repeatedly on the same `Configuration`.
+- Use a fresh `depResults` map per build pass: the map memoises each configuration's result, and each external dependency's fetch and build, for one pass, so reusing it skips rebuilds and pulls you meant to observe. `Build()` itself is safe to call repeatedly on the same `Configuration`.
 - Set `cfg.path` before calling `GetInterfacesAndImplementations`, and spell `cFiles` as absolute paths: both resolve against the working directory, not the configuration.
 
 A bug fix ships with a test that fails before the fix and passes after it, in its own `test:` commit when it documents the behaviour on its own.