refactor: use Timestamp{Nanosecond} for time axes instead of UnixTime

[Beforerr]

Time coordinates are now Durations.Timestamp{Nanosecond} (stdlib Dates.Timestamp on Julia 1.14+), still a zero-copy reinterpret of numpy datetime64[ns]. Drops UnixTimes.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[github-actions]

Benchmark Results (Julia v1)

Time benchmarks

	main	66e6385...	main / 66e6385...
DimArray	16.6 ± 1.8 μs	17.7 ± 2.2 μs	0.939 ± 0.15
DimArray (direct)	22.6 ± 1.8 μs	24.1 ± 2.6 μs	0.939 ± 0.13
get_data	4.72 ± 0.46 μs	4.55 ± 0.37 μs	1.04 ± 0.13
get_data (collect)	6.88 ± 2.8 μs	7.02 ± 2.6 μs	0.98 ± 0.54
time_to_load	3.17 ± 0.058 s	3.41 ± 0.024 s	0.929 ± 0.018

Memory benchmarks

	main	66e6385...	main / 66e6385...
DimArray	0.052 k allocs: 2.28 kB	0.052 k allocs: 2.28 kB	1
DimArray (direct)	0.084 k allocs: 3.45 kB	0.084 k allocs: 3.45 kB	1
get_data	0.032 k allocs: 1.17 kB	0.032 k allocs: 1.17 kB	1
get_data (collect)	0.035 k allocs: 6.87 kB	0.035 k allocs: 6.87 kB	1
time_to_load	0.194 k allocs: 11.2 kB	0.194 k allocs: 11.2 kB	1

[Beforerr]

Merged to main as 8779850 (rebased onto SpaceDataModel 0.4 compat).