0.4.0 --- repin the C library, which no longer answers a question with another one's answer

openkal-musl 0.7.0. Nothing in this package changes: it names the C library and
the C library named the specification, so what moves is one line.

What the consumers of this package get is measured rather than asserted. The
same `std::filesystem` program, run over this stack and over the host's own
toolchain, now agrees row for row where it did not:

  stat on a node naming another        was a link, is what the name refers to
  is_regular_file on the same          was false for bytes it could read
  file_size on the same                was refused
  exists on one naming nothing         was true
  is_directory in an enumeration       was false for one naming a directory
  copy of a tree holding one           was ENOSYS; one such node made a whole
                                       tree uncopyable
  readlink                             was ENOSYS
  equivalent of two different files    was TRUE, with no error

and `signal(SIGABRT, …)`, which any terminal library, death test or crash
reporter performs before it does anything else, no longer ends the program.

Author: speak-agent

mcpp.toml

@@ -1,7 +1,7 @@
 [package]
 namespace   = "mcpplibs"
 name        = "openkal-llvm-runtime"
-version     = "0.3.1"
+version     = "0.4.0"
 description = "LLVM's C++ runtime libraries — libc++, libc++abi and libunwind — configured for openkal-musl rather than for a host C library."
 license     = "Apache-2.0"
 authors     = ["mcpplibs"]
@@ -208,7 +208,7 @@ sources = [
 cflags = ["-DDISABLE_AARCH64_FMV=1"]
 
 [dependencies]
-openkal-musl = "0.6.0"
+openkal-musl = "0.7.0"
 
 [build]
 cxx_standard = "c++23"