ok    1  %% after a conversion is one literal % (was x%%)
ok    2  %% at the start, mid-text, after %d
ok    3  %%%% = %%, several %%, trailing %%
ok    4  %% between two conversions
ok    5  %10s / %-10s
ok    6  %3s longer than the width; %10s of empty
ok    7  %5d / %-5d / %5i
ok    8  %08d zero pad, sign first
ok    9  %5d negative; - overrides 0
ok    10  width smaller than the digits; %*d still works
ok    11  two padded conversions with tails
ok    12  %.2m positive / negative
ok    13  %12.2m / %-12.2m
ok    14  %+.2m forced sign
ok    15  half-even ties
ok    16  negative zero, carry, .15
ok    17  width-only, integer-valued reals
ok    18  bare %m / %f (forward str$ writer)
ok    19  %r / %f with precision unchanged (snprintf path)
ok    20  a WIDE value keeps the m_wide_apply path
ok    21  pure fraction; tie at the third decimal
ok    22  %T with lower-case list words
ok    23  %T apostrophe rules
ok    24  plural table via binary search: 
ok    25  %t items via the name-length table
ok    26  escapes resolved at parse time
ok    27  the same cached format with different arguments
ok    28  300 distinct formats (more than the 256-entry cache)
ok    29  the first format again after eviction
ok    30  a format over 16K raises
ok    31  empty and literal-only formats

passed:  31   failed:  0 
