GH-1153 Rework delay system with per-entry Instant TTL

[imDMK]

Description

This PR refactors the delay handling to eliminate cache-level TTL conflicts.

• Introduced GuavaDelay with shared logic on top of Guava Cache
• Removed expireAfterWrite to ensure TTL is controlled per entry via Instant
• Added separate interfaces: DefaultDelay (with predefined duration) and ExplicitDelay (explicit duration required)
• Implemented GuavaDefaultDelay and GuavaExplicitDelay
• Added Delay factory for convenient instance creation
• Clarified contract and documentation

Result: cleaner API, safer usage, no premature cache evictions.
Fixes #1153

Type of change

• Bug fix (non-breaking change which fixes an issue)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• This change requires a documentation update

How Has This Been Tested?

• Localhost server

[gemini-code-assist]

Code Review

This pull request refactors the delay system to use per-entry TTLs, which is a significant improvement for avoiding cache-level conflicts. The introduction of DefaultDelay and ExplicitDelay interfaces, along with a factory, creates a much cleaner and safer API. My review highlights a critical issue where expireAfterWrite was unintentionally reintroduced in GuavaDefaultDelay, undermining the primary goal of this refactoring. I have also included some suggestions to improve the API design by reducing code duplication and better aligning the interface documentation with its behavior, which would enhance flexibility and maintainability.

[P1otrulla]

Good job i like it but i'm thinking about overenginering at this case - maybe Im wrong

[imDMK]

/gemini review

[gemini-code-assist]

Code Review

This pull request provides a solid refactoring of the delay system, moving from a cache-level TTL to a per-entry TTL managed by Instant. This is a great improvement for correctness and prevents premature evictions. The introduction of DefaultDelay and ExplicitDelay interfaces, along with a factory class, results in a much cleaner and more robust API. The implementation is well-structured, especially the shared logic in GuavaDelay. I've identified a few minor inconsistencies in handling expiration edge cases that could be addressed to further improve the robustness of the new system. Overall, this is a high-quality contribution.

[Jakubk15]

Looks solid to me. Shout-out for the good documentation of the API.

[Jakubk15]

Rename to FixedDelay. Sounds better imo