Merge bitcoin#29846: guix: replace GCC unaligned VMOV patch with binutils patch

a0dc2eb guix: replace GCC unaligned VMOV patch with binutils patch (fanquake)

Pull request description:

  Rather than invasively patching GCC, given we have binutils 2.38 available, we can patch it to flip the default for
  `-muse-unaligned-vector-move`.

  A 1 line binutils patch, is much more maintainable than the ~300 line patch into GCC. It's also a slight inprovement in regards to patching out ualigned instructions in the release binaries. For comparison:
  Master:
  ```bash
  objdump -D bin/*.exe | rg "vmova|vmovdqa|vmovaps|vmovapd|vmovdqa64|vmovdqa32"
  141b8be20: c5 f8 28 1a                 vmovaps(%rdx), %xmm3
  1420564b3: c5 79 29 36                 vmovapd%xmm14, (%rsi)
  1403060f3: c5 79 29 36                 vmovapd%xmm14, (%rsi)
  140792b13: c5 79 29 36                 vmovapd%xmm14, (%rsi)
  140cb0693: c5 79 29 36                 vmovapd%xmm14, (%rsi)
  1415ea0f3: c5 79 29 36                 vmovapd%xmm14, (%rsi)
  ```

  This PR:
  ```bash
  objdump -D bin/*.exe | rg "vmova|vmovdqa|vmovaps|vmovapd|vmovdqa64|vmovdqa32"
  141b8be20: c5 f8 28 1a                 vmovaps(%rdx), %xmm3
  1420564b3: c5 79 29 36                 vmovapd%xmm14, (%rsi)
  1403060f3: c5 79 29 36                 vmovapd%xmm14, (%rsi)
  140792b13: c5 79 29 36                 vmovapd%xmm14, (%rsi)
  140cb0693: c5 79 29 36                 vmovapd%xmm14, (%rsi)
  ```

ACKs for top commit:
  laanwj:
    Code review ACK a0dc2eb

Tree-SHA512: b3b6dcd2efaaa825d32c768302651d26a120a3e47b93fafb862a1884ff68fd96edb42ea9bc9974c005c8f5a1d15c217deec0ed462cc4a3365cab1bad5a0b5fef

Author: fanquake

contrib/guix/manifest.scm

@@ -110,12 +110,15 @@ desirable for building Bitcoin Core release binaries."
 
 (define (gcc-mingw-patches gcc)
   (package-with-extra-patches gcc
-    (search-our-patches "gcc-remap-guix-store.patch"
-                        "vmov-alignment.patch")))
+    (search-our-patches "gcc-remap-guix-store.patch")))
+
+(define (binutils-mingw-patches binutils)
+  (package-with-extra-patches binutils
+    (search-our-patches "binutils-unaligned-default.patch")))
 
 (define (make-mingw-pthreads-cross-toolchain target)
   "Create a cross-compilation toolchain package for TARGET"
-  (let* ((xbinutils (cross-binutils target))
+  (let* ((xbinutils (binutils-mingw-patches (cross-binutils target)))
          (pthreads-xlibc mingw-w64-x86_64-winpthreads)
          (pthreads-xgcc (cross-gcc target
                                     #:xgcc (gcc-mingw-patches mingw-w64-base-gcc)

contrib/guix/patches/binutils-unaligned-default.patch

@@ -0,0 +1,22 @@
+commit 6537181f59ed186a341db621812a6bc35e22eaf6
+Author: fanquake <[email protected]>
+Date:   Wed Apr 10 12:15:52 2024 +0200
+
+    build: turn on -muse-unaligned-vector-move by default
+
+    This allows us to avoid (more invasively) patching GCC, to avoid
+    unaligned instruction use.
+
+diff --git a/gas/config/tc-i386.c b/gas/config/tc-i386.c
+index e0632681477..14a9653abdf 100644
+--- a/gas/config/tc-i386.c
++++ b/gas/config/tc-i386.c
+@@ -801,7 +801,7 @@ static unsigned int no_cond_jump_promotion = 0;
+ static unsigned int sse2avx;
+ 
+ /* Encode aligned vector move as unaligned vector move.  */
+-static unsigned int use_unaligned_vector_move;
++static unsigned int use_unaligned_vector_move = 1;
+ 
+ /* Encode scalar AVX instructions with specific vector length.  */
+ static enum