feat: Enable SIMD optimizations

* Add x86_64-microarch-level as a 'build' requirement.
   - microarch_level 4 not supported yet so only add level 1 to 3.
* Set the build number based on the microarch_level.
* Add conda_build_config.yaml with microarch_level.
* c.f. https://conda-forge.org/docs/maintainer/knowledge_base/#microarch
* Bump build number.

recipe/conda_build_config.yaml

@@ -0,0 +1,3 @@
+microarch_level:
+  - 1
+  - 3  # [unix and x86_64]

recipe/meta.yaml

@@ -1,5 +1,6 @@
 {% set name = "boost-histogram" %}
 {% set version = "1.5.0" %}
+{% set build = 2 %}
 
 package:
   name: {{ name|lower }}
@@ -10,7 +11,12 @@ source:
   sha256: 0623f010e6c52e5d018767723959686090db07fc30f0d1d8475b5d663c5ddb2c
 
 build:
-  number: 1
+  # Prioritize builds with a higher microarch level.
+  # microarch_level 4 not supported yet.
+  # c.f. https://github.com/conda-forge/microarch-level-feedstock/issues/5
+  number: {{ build }}          # [not (unix and x86_64)]
+  number: {{ build + 100 }}    # [unix and x86_64 and microarch_level == 1]
+  number: {{ build + 300 }}    # [unix and x86_64 and microarch_level == 3]
   script:
     - export CMAKE_BUILD_PARALLEL_LEVEL=${CPU_COUNT}  # [not (linux and ppc64le)]
     - {{ PYTHON }} -m pip install . -v
@@ -19,6 +25,7 @@ requirements:
   build:
     - python                                 # [build_platform != target_platform]
     - cross-python_{{ target_platform }}     # [build_platform != target_platform]
+    - x86_64-microarch-level {{ microarch_level }}  # [unix and x86_64]
     - {{ compiler('cxx') }}
     - {{ stdlib("c") }}
     - cmake