share/mm: synced with {mm v5.0.0-0-g81b3036}

meta: bumped the version to {1.12.5}

cmake: added the cuda bindings to the build

github: the {cmake} workflow can now be triggered manually

github: build {wheels} on {ubuntu-22.04}

pyre.h5: removed some left over debugging logic

merlin.cli: {lib} now asks {library} for its {assets} in the new way

cmake: added the new {journal} test cases and their cleanup logic

pyre::grid: added {order_t} bindings; unlike their c++ counter parts,…

… the python {order} classes don't get any {rep_t} algebra; i think the c++ implementation is overdue for some restructuring

merlin.components: renamed {authority} to {visitor}