feat: Improve intel MKL shared library build process using docker

[sudiptob2]

MathNet.Numerics supports Intel MKL for CPU acceleration. On Linux, MKL support relies on the native provider MathNet.Numerics.MKL.Linux. Currently, this provider is not bundled with MathNet.Numerics due to size constraints. Instead, it is available as a separate NuGet package. However:

• The NuGet package is outdated. Please take a look at this discussion.
• It is incompatible with the latest builds from the master branch.

To work with MKL, users must build the provider manually, which involves compiling the Intel shared libraries using the mkl_build.sh script. This script is outdated and depends on pre-installed host machine dependencies.

To streamline the process, in this PR following is proposed:

1. Updating the build script for Ubuntu 24.04.
2. Adding a Dockerfile to build the shared libraries inside a container, eliminating host dependency requirements.
3. Improving documentation on how to use the script and Dockerfile.