fix: replace static buffer allocation on growth

[smira]

Use slices.Grow and follow whatever algorithm it uses internally to pick the next size of the buffer.

So the growth might not be exactly what we asked for, but it should be optimal from Go runtime point of view.

[DmitriyMV]

🆒

[smira]

No real changes with the benchmark:

name                     old time/op    new time/op    delta
Write/defaults-32           152ns ± 3%     153ns ± 5%    ~     (p=0.841 n=5+5)
Write/growing_buffer-32     154ns ± 5%     152ns ± 3%    ~     (p=0.690 n=5+5)
Write/compression-32        425ns ± 3%     423ns ± 2%    ~     (p=1.000 n=5+5)
Read/defaults-32           11.9µs ± 3%    11.9µs ± 2%    ~     (p=0.841 n=5+5)
Read/growing_buffer-32     11.7µs ± 3%    11.9µs ± 2%    ~     (p=0.310 n=5+5)
Read/compression-32         338µs ± 1%     342µs ± 2%    ~     (p=0.151 n=5+5)

name                     old alloc/op   new alloc/op   delta
Write/defaults-32           0.00B          0.00B         ~     (all equal)
Write/growing_buffer-32     0.00B          0.00B         ~     (all equal)
Write/compression-32       1.02kB ± 0%    1.02kB ± 0%    ~     (all equal)
Read/defaults-32             128B ± 0%      128B ± 0%    ~     (all equal)
Read/growing_buffer-32       128B ± 0%      128B ± 0%    ~     (all equal)
Read/compression-32         139kB ± 0%     139kB ± 0%  -0.00%  (p=0.016 n=5+4)

name                     old allocs/op  new allocs/op  delta
Write/defaults-32            0.00           0.00         ~     (all equal)
Write/growing_buffer-32      0.00           0.00         ~     (all equal)
Write/compression-32         0.00           0.00         ~     (all equal)
Read/defaults-32             1.00 ± 0%      1.00 ± 0%    ~     (all equal)
Read/growing_buffer-32       1.00 ± 0%      1.00 ± 0%    ~     (all equal)
Read/compression-32          2.00 ± 0%      2.00 ± 0%    ~     (all equal)

[smira]

/m