Simplify controller manager namespacing

[stephanie-eng]

Description

This PR attempts to simplify the logic with setting a namespace (stored in the ns_ variable) for Ros2Control Manager.
My understanding of the original logic:

• ns_ can be set via constructor, but its value is ignored and replaced by the parameter value.
• If that parameter doesn't exist, we default the parameter to "/" and also set ns_ to match.
• In the case where ns_ is not set via constructor, we try to get it from a parameter.
• If that parameter doesn't exist, ns_ remains an empty string.

It sounds like the desired behaviour would be to set ns_ to the value of the ros_control_namespace parameter, if it exists, or otherwise default to "/", without considering the original value of ns_.

This PR also adds a deprecation notice to setting ns_ via constructor, since this value is ignored anyway. I've deprecated it instead of removing it outright to be safe, but I'm not sure that this is used anywhere, given that this is used as a plugin and thus should be default constructed.

[codecov]

Codecov Report

Patch coverage has no change and project coverage change: -0.01 ⚠️

Comparison is base (b98bb6b) 50.53% compared to head (2dd1d73) 50.51%.

❗ Current head 2dd1d73 differs from pull request most recent head 0ee7934. Consider uploading reports for the commit 0ee7934 to get more accurate results

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2210      +/-   ##
==========================================
- Coverage   50.53%   50.51%   -0.01%     
==========================================
  Files         386      386              
  Lines       31736    31736              
==========================================
- Hits        16034    16029       -5     
- Misses      15702    15707       +5     

see 6 files with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[sea-bass]

Regardless of whether the logic is equivalent, I agree that we should stick to a single way of specifying this parameter.