Use Ulid instead of Guid for better PERF

[hishamco]

The benchmarks show that the Ulid is faster https://github.com/Cysharp/Ulid

FYI .NET 9.0 introduces Guid.CreateVersion7() but according to my findings Ulid is still the fastest

Method	Mean	Error	StdDev
NewGuidV4	45.01 ns	0.119 ns	0.106 ns
NewGuidV7	82.69 ns	0.821 ns	0.768 ns
NewUlid	29.90 ns	0.098 ns	0.092 ns
NewUlidToGuid	31.12 ns	0.185 ns	0.164 ns

[hishamco]

Also, we could introduce IGuidGenerator in OC.Infrastructure.Abstraction that has the default implementation, so anyone else can replace the default implementation easily without touch the OC code-base

[sebastienros]

For the sake of the reasoning, we are talking 50ns in the best case. Which is 20 times less than a micro second (not milli, micro). For usages which are not in hot paths. And I can only see drawbacks, like introducing another external dependency, with potentially security implications (is it really unique?), and serialization concerns (will STJ support it like standard dotnet Guids?), and deviating from the standard dotnet libs.

Note that the author is very well trusted and we already use ZString or MessagePack.

[hishamco]

And I can only see drawbacks, like introducing another external dependency

Right, but we already used external dependency in many places for example using ZString

Note that the author is very well trusted and we already use ZString or MessagePack.

Exactly we already used ZString

I'm wondering about the STJ serialization, if we intend to use such a thing I might spend some time testing the serialization, but I think it's doable

[github-actions]

This pull request has merge conflicts. Please resolve those before requesting a review.

[github-actions]

This pull request has merge conflicts. Please resolve those before requesting a review.

[github-actions]

This pull request has merge conflicts. Please resolve those before requesting a review.

[github-actions]

This pull request has merge conflicts. Please resolve those before requesting a review.