🏄 `prepare_spectra()`: Improving speed while maintaining reasonable memory consumption

[philipp-baumann]

addresses part of #52

[philipp-baumann]

My conclusion is that at least in Reduce context, append() instead of c() is neither faster nor better in terms of memory (needs also more frequent calls to Garbage Collector.

before (current master) : c3e4c2d38594574f38de48283e6e181ddfda8dfb

r$> library("opusreader2")
    file <- opus_file()

r$> bnch <- bench::mark(
      data <- read_opus(dsn = file)
    )

r$> bnch
# A tibble: 1 × 13
  expression                         min   median `itr/sec` mem_alloc `gc/sec` n_itr  n_gc total_time result            memory               time            gc               
  <bch:expr>                    <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl> <int> <dbl>   <bch:tm> <list>            <list>               <list>          <list>           
1 data <- read_opus(dsn = file)   17.5ms   17.9ms      54.4    1.73MB     172.     6    19      110ms <named list [20]> <Rprofmem [504 × 3]> <bench_tm [25]> <tibble [25 × 3]>

after (this PR)

r$> library("opusreader2")
    file <- opus_file()

r$> bnch <- bench::mark(
      data <- read_opus(dsn = file)
    )
Warning message:
Some expressions had a GC in every iteration; so filtering is disabled. 

r$> bnch
# A tibble: 1 × 13
  expression                         min   median `itr/sec` mem_alloc `gc/sec` n_itr  n_gc total_time result            memory               time            gc               
  <bch:expr>                    <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl> <int> <dbl>   <bch:tm> <list>            <list>               <list>          <list>           
1 data <- read_opus(dsn = file)   19.4ms     20ms      46.7    2.07MB     52.6    24    27      514ms <named list [20]> <Rprofmem [577 × 3]> <bench_tm [24]> <tibble [24 × 3]>

[philipp-baumann]

closing because this PR with append has no advantages.