Update the AWS, GCP and Azure deployment pages #1790

The three cloud pages under pages/deployment/environments/ (aws.mdx, gcp.mdx, azure.mdx) are copies of one template and have drifted from the rest of the docs. Link and anchor fixes are in #1789. This issue tracks the content.

Out of date

• MAGE on a native install. The pages say MAGE has to be built from source. Prebuilt memgraph-mage .deb and .rpm packages now exist (/advanced-algorithms/install-mage).
• ARM support. The pages say CentOS and Fedora don't run on ARM. The download page now has distro-agnostic aarch64 .deb and .rpm packages. The seven per-distro install links reflect the old packaging too.
• The Docker command doesn't match the text. The text says to use Docker because it includes MAGE, but the command runs memgraph/memgraph:latest, which doesn't. The install guide recommends memgraph/memgraph-mage and also publishes port 7444.
• HA ports. The pages say coordinators need only port 12000. HA best practices say coordinators also need --management-port (10000). Present the ports as the example values from best practices, not fixed requirements.
• HA coverage is uneven. Only AWS has an HA section, and it ends mid-sentence ("This provides:" with nothing after it). Azure doesn't link the AKS guide in helm-charts (charts/memgraph-high-availability/aks). The charts repo has no GCP/GKE guide.

Cleanup

• Azure: remove the stray bare URL in "Setup the storage".
• Fix typos ("instance***", "Magnetic storage can also be used" split over two lines).

Structure

• Consider merging the three pages into one shared cloud VM page with a short section per provider, so the content can't drift three ways again. Keep redirects from the current paths.