fix: ensure file systems with matching roleArns are registered correctly TDE-1268

[blacha]

Motivation

file systems are only created for unique roleArns so FileSystemCreated event is not fired twice if two configuration objects have the same roleArn,

give a confiugration with two buckets both using roleA on a service that has a default role roleDefault

s3://foo/ - roleA
s3://bar/ - roleA

requests to

fs.read("s3://foo"); // tries roleDefault then uses roleA
fs.read("s3://foo"); // uses cached roleA

depending on the order of reads the default role may be used far too often

fs.read("s3://foo") // tries roleDefault then uses roleA
fs.read("s3://bar") // tries roleDefault then uses roleA
fs.read("s3://bar") // tries roleDefault then uses roleA

after this change

fs.read("s3://foo") // tries roleDefault then uses roleA
fs.read("s3://bar") // tries roleDefault then uses roleA
fs.read("s3://bar") // uses cached roleA

Modification

hook the file system finder when it needs to find a new file system use that file system to register onto fsa

Checklist

If not applicable, provide explanation of why.

• Tests updated
• Docs updated
• Issue linked in Title