Use newtype for Index/(Un)Signed on GHC 9.4+

[christiaanb]

Implements #2525

Most benchmarks were within 1-3% of each other. The outliers, in favour of newtype, were:

strict field:

benchmarking Signed/+ 3*WORD_SIZE_IN_BITS
time                 120.3 ns   (119.8 ns .. 120.8 ns)
                     1.000 R²   (1.000 R² .. 1.000 R²)
mean                 120.8 ns   (120.4 ns .. 121.3 ns)
std dev              1.520 ns   (1.333 ns .. 1.729 ns)
variance introduced by outliers: 13% (moderately inflated)

newtype:

benchmarking Signed/+ 3*WORD_SIZE_IN_BITS
time                 106.3 ns   (106.3 ns .. 106.4 ns)
                     1.000 R²   (1.000 R² .. 1.000 R²)
mean                 106.3 ns   (106.3 ns .. 106.4 ns)
std dev              133.3 ps   (107.9 ps .. 171.5 ps)

strict field:

benchmarking Signed/- 3*WORD_SIZE_IN_BITS
time                 95.15 ns   (95.11 ns .. 95.20 ns)
                     1.000 R²   (1.000 R² .. 1.000 R²)
mean                 95.14 ns   (95.10 ns .. 95.19 ns)
std dev              148.2 ps   (123.6 ps .. 183.9 ps)

newtype:

benchmarking Signed/- 3*WORD_SIZE_IN_BITS
time                 89.93 ns   (89.04 ns .. 91.70 ns)
                     0.999 R²   (0.997 R² .. 1.000 R²)
mean                 89.35 ns   (89.05 ns .. 90.99 ns)
std dev              1.787 ns   (60.20 ps .. 4.337 ns)
variance introduced by outliers: 28% (moderately inflated)

strict field:

benchmarking Signed/.&. WORD_SIZE_IN_BITS
time                 87.26 ns   (87.24 ns .. 87.29 ns)
                     1.000 R²   (1.000 R² .. 1.000 R²)
mean                 87.24 ns   (87.21 ns .. 87.27 ns)
std dev              89.59 ps   (71.91 ps .. 108.3 ps)

newtype:

benchmarking Signed/.&. WORD_SIZE_IN_BITS
time                 76.58 ns   (76.55 ns .. 76.61 ns)
                     1.000 R²   (1.000 R² .. 1.000 R²)
mean                 76.60 ns   (76.57 ns .. 76.62 ns)
std dev              92.21 ps   (67.79 ps .. 137.1 ps)

strict field:

benchmarking Signed/.|. WORD_SIZE_IN_BITS
time                 89.61 ns   (89.56 ns .. 89.65 ns)
                     1.000 R²   (1.000 R² .. 1.000 R²)
mean                 89.59 ns   (89.55 ns .. 89.63 ns)
std dev              147.2 ps   (109.2 ps .. 208.7 ps)

newtype:

benchmarking Signed/.|. WORD_SIZE_IN_BITS
time                 73.43 ns   (72.92 ns .. 74.35 ns)
                     1.000 R²   (0.999 R² .. 1.000 R²)
mean                 73.07 ns   (72.89 ns .. 74.16 ns)
std dev              976.2 ps   (352.4 ps .. 2.453 ns)
variance introduced by outliers: 15% (moderately inflated)

strict field:

benchmarking Unsigned/.|. 3*WORD_SIZE_IN_BITS
time                 62.10 ns   (62.08 ns .. 62.12 ns)
                     1.000 R²   (1.000 R² .. 1.000 R²)
mean                 62.10 ns   (62.09 ns .. 62.11 ns)
std dev              37.36 ps   (30.95 ps .. 46.13 ps)

newtype:

benchmarking Unsigned/.|. 3*WORD_SIZE_IN_BITS
time                 48.72 ns   (48.71 ns .. 48.72 ns)
                     1.000 R²   (1.000 R² .. 1.000 R²)
mean                 48.72 ns   (48.71 ns .. 48.72 ns)
std dev              11.83 ps   (9.698 ps .. 14.92 ps)

Still TODO:

• [x] Write a changelog entry (see changelog/README.md)
• [x] Check copyright notices are up to date in edited files