Fix performance of DPF vector

[cbellot000]

closing #2201

DPF vector became slow because they were always checking if the modification of the vector by

1. making a deep copy of the vector
2. comparing the whole vector

Why performance changed between 0.13.6 and master?
Before 0.13.7.dev, it was hard coded that in process dpf vector wouldn't "check modification" and though bypassed dpf server logic. 0.13.7.dev is now relying on DPF server logic for updating or not dpf vector.

Proposed fixes

• DPFVector are now always "committed" to dpf server whether they have been modified or not (as checking for modification is very slow)
• Performance is highly improved for server version checks by caching previous checks

Benchmark
Test is taken for #2201, changing params are:

• pydpf-core version
• server types: in process/gRPC
• vector operations: data is only "gotten", or data is "gotten" and "set"

Server type	Vec operation	0.13.6	0.13.7.dev
In Process	Get	0.009 s	0.008 s
In Process	Get/Set	0.012 s	0.011 s
gRPC	Get	6.608 s	3.773 s
gRPC	Get/Set	8.707 s	3.707 s

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 84.03%. Comparing base (eb9035f) to head (748b4df).
Report is 1 commits behind head on master.

✅ All tests successful. No failed tests found.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2249      +/-   ##
==========================================
+ Coverage   84.01%   84.03%   +0.02%     
==========================================
  Files          90       90              
  Lines       10308    10316       +8     
==========================================
+ Hits         8660     8669       +9     
+ Misses       1648     1647       -1