fifocache: refine key hash func and optimize evict

[reusee]

What type of PR is this?

• [ ] API-change
• [ ] BUG
• [x] Improvement
• [ ] Documentation
• [ ] Feature
• [ ] Test and CI
• [ ] Code Refactoring

Which issue(s) this PR fixes:

issue #19453

What this PR does / why we need it:

refine key shard function and optimize evict benchstat:

goos: linux
goarch: amd64
pkg: github.com/matrixorigin/matrixone/pkg/fileservice/fifocache
cpu: AMD Ryzen 9 7900 12-Core Processor             
                      │      b1      │                  b2                  │
                      │    sec/op    │    sec/op     vs base                │
SequentialSet-24        130.7n ± ∞ ¹   137.0n ± ∞ ¹       ~ (p=0.100 n=3) ²
ParallelSet-24          27.22n ± ∞ ¹   27.57n ± ∞ ¹       ~ (p=0.100 n=3) ²
Get-24                  19.43n ± ∞ ¹   19.86n ± ∞ ¹       ~ (p=0.200 n=3) ²
ParallelGet-24          5.083n ± ∞ ¹   5.169n ± ∞ ¹       ~ (p=0.400 n=3) ²
ParallelGetOrSet-24     20.35n ± ∞ ¹   19.89n ± ∞ ¹       ~ (p=0.400 n=3) ²
ParallelEvict-24        226.4n ± ∞ ¹   140.0n ± ∞ ¹       ~ (p=0.100 n=3) ²
EnsureNBytesAndSet-24   34.11n ± ∞ ¹   35.13n ± ∞ ¹       ~ (p=0.100 n=3) ²
geomean                 34.24n         32.46n        -5.22%