fix: Use a single timestamp in server_info

[snowyukitty]

server_info sampled system_clock once when constructing InfoSection::time
and again when calculating validated_ledger.age. If those samples straddled a
whole-second boundary, the exact-age assertions could fail intermittently.

This change reuses the timestamp already stored in output.info.time for the
age calculation. The clock remains system_clock by default; making it
injectable lets the tests use fixed absolute ledger times and verify that
successful requests sample it exactly once. A regression test also preserves
the existing zero clamp for future close times.

Fix #2353

Testing

• RPCServerInfoHandlerTest.*: 12 passed; repeated 100 times (1,200 executions)
• Adjacent RPC handler and ledger tests: 75 passed
• Full unit suite: 3,294 passed, 1 disabled
• GCC 15.2 Release with -Werror; affected Clang 19 translation units
• clang-format 22.1.8, clang-tidy, C++ pre-commit hooks, codespell, and Doxygen
  1.14.0

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[godexsoft]

Thank you, this is great! just one nit and i think this is good to go 👍

[godexsoft]

Nice 👍

[mathbunnyru]

@snowyukitty I'm sorry for a slow PR review here, will you be able to take a look here?

[snowyukitty]

Thanks — all four addressed in d02e270.

• ClockType is constrained by a new util::SomeSystemClock concept in util/Concepts.hpp instead of being described in a doc comment.
• The test clock moved out of ServerInfoTests.cpp into its own header, tests/common/util/TestConstantClock.hpp, so other suites can use it.
• It is a class: the fixed instant (kNowUnix, kNow) is public on it, the counter is private, and it exposes callCount() and resetCounter(). The fixture calls TestConstantClock::resetCounter() instead of assigning to a public member.

Rebased onto your develop merge. clio_tests is 3401 passed, RPCServerInfoHandlerTest.* 12 passed, and clang-format is clean on the four files.

[mathbunnyru]

@snowyukitty, your PR was failing in clang-tidy; I fixed it.

May I ask you to sign your commit?
You can reset the diff on top of the latest develop (please make sure to update it), and make it a single signed commit.