perf: reuse allocs in serialization

[Vonr]

main branch version of #50
vs main:

read/complex_player     time:   [12.801 µs 12.988 µs 13.187 µs]
                        thrpt:  [244.45 MiB/s 248.18 MiB/s 251.82 MiB/s]
                 change:
                        time:   [-3.9509% -1.1581% +1.9716%] (p = 0.43 > 0.05)
                        thrpt:  [-1.9335% +1.1717% +4.1134%]
                        No change in performance detected.
Found 10 outliers among 100 measurements (10.00%)
  9 (9.00%) low mild
  1 (1.00%) high mild

read/chunk              time:   [39.183 µs 39.850 µs 40.602 µs]
                        thrpt:  [1.0168 GiB/s 1.0361 GiB/s 1.0537 GiB/s]
                 change:
                        time:   [-2.6364% +1.0505% +4.8395%] (p = 0.57 > 0.05)
                        thrpt:  [-4.6161% -1.0396% +2.7078%]
                        No change in performance detected.
Found 1 outliers among 100 measurements (1.00%)
  1 (1.00%) high mild

     Running benches/write.rs (target/release/deps/write-c5d68cceaef7608a)
write/complex_player    time:   [4.4602 µs 4.4801 µs 4.4963 µs]
                        thrpt:  [716.91 MiB/s 719.49 MiB/s 722.70 MiB/s]
                 change:
                        time:   [-82.990% -82.743% -82.497%] (p = 0.00 < 0.05)
                        thrpt:  [+471.34% +479.46% +487.88%]
                        Performance has improved.

write/chunk             time:   [20.042 µs 20.266 µs 20.484 µs]
                        thrpt:  [2.0155 GiB/s 2.0372 GiB/s 2.0600 GiB/s]
                 change:
                        time:   [-75.588% -75.030% -74.443%] (p = 0.00 < 0.05)
                        thrpt:  [+291.28% +300.47% +309.64%]
                        Performance has improved.
Found 15 outliers among 100 measurements (15.00%)
  15 (15.00%) low mild

[SzczurekYT]

All good, I only moved serialize_str_into to utils.rs to match string deserialization being there too