Fixes #161.

Nixpkgs' stdenv disables dependency tracking by default.  That makes
sense for one-time builds, but in an interactive environment we expect
repeated "make" invocations to do the right thing.

This allows scripts to distinguish between a real build and a Nix
shell.

Setting $NIX_STORE causes the purity checks in gcc/ld-wrapper to kick
in, so that's why we unset $NIX_ENFORCE_PURITY.

This causes the environment to be (almost) cleared, thus giving a
shell that more closely resembled the actual Nix derivation.

This reverts commit 69b8f998.

The timeout should be enforced remotely.  Otherwise, if the garbage
collector is running either locally or remotely, if will block the
build or closure copying for some time.  If the garbage collector
takes too long, the build may time out, which is not what we want.
Also, on heavily loaded systems, copying large paths to and from the
remote machine can take a long time, also potentially resulting in a
timeout.

Fixes #113.
Fixes #131.

This ensures that not just environment variables are set, but also
shell functions such as unpackPhase, configurePhase and so on.

Issue NixOS/hydra#102.

For instance, it's pointless to keep copy-from-other-stores running if
there are no other stores, or download-using-manifests if there are no
manifests.  This also speeds things up because we don't send queries
to those substituters.

Thus passing ‘--option debug-subst 1’ allows daemon users to turn on
debug info and see what the substituter is doing.

Previously, if a binary cache is hanging/unreachable/slow,
download-from-binary-cache.pl would also hang without any indication
to the user.  Now, if fetching a URL takes more than 5 seconds, it
will print a message to that effect.

Fixes the error "DBD::SQLite::db do failed: column url is not unique".

Amazon S3 returns HTTP status code 403 if a file doesn't exist and the
user has no permission to list the contents of the bucket.  So treat
it as 404 (meaning it's cached in the NARExistence table).

The "$UID != 0" makes no sense: if the local side has write access to
the Nix store (which is always the case) then it doesn't matter if
we're root - we can import unsigned paths either way.

Fixes #120.

Reported by Pablo Costa.

Otherwise it will set the parent's stdin to non-blocking mode, causing
the subsequent read of the set of inputs/outputs to fail randomly.
That's insane.

Before selecting a machine, build-remote.pl will try to run the
command "nix-builds-inhibited" on the machine.  If this command exists
and returns a 0 exit code, then the machine won't be used.  It's up to
the user to provide this command, but it would typically be a script
that checks whether there is enough disk space and whether the load is
not too high.

Don't pass --timeout / --max-silent-time to the remote builder.
Instead, let the local Nix process terminate the build if it exceeds a
timeout.  The remote builder will be killed as a side-effect.  This
gives better error reporting (since the timeout message from the
remote side wasn't properly propagated) and handles non-Nix problems
like SSH hangs.

This allows providing additional binary caches, useful in scripts like
Hydra's build reproduction scripts, in particular because untrusted
caches are ignored.

Problem noticed by niksnut.

Based on https://github.com/NixOS/nix/pull/6 from shlevy

For instance, this prevents paths from being deleted that are in use
by a "nix-build --run-env" session.

Reported by Peter Simons.