Previously NIL could be passed as the :START argument to POSITION-IF,
which resulted in signaling an implementation-specific error.

Issue reported by saltrocklamp[m] on Libera #commonlisp.

Added new test cases for positive numbers with + prefixed and provided
fix for floating point numbers starting with "+" sign followed
immediately by "." (e.g. +.14)

This ensures it is copied by bundle-op

Fixes issue #6.

Initially when I put docstrings in the inlined functions, I was sure I
got some warning about how the docstring was inserted into the code so
it was being optimized out or something. I must been doing something
weird because it is no longer happening.

There are checks to make sure that if a number is not in decimal, it
does not have an exponent marker. Unfortunately, the code that detects
an exponent marker, will only do so if the number is not a decimal
number. This means numbers in the number #x5l0 the 'l' would not be
considered an exponent marker. The good news is that any of these
cases would have eventually been caught by parse-integer, since it
would be passed the exponent marker, signaling an error because it is
not a digit in the given radix.

To fix this, the code the detects an exponent marker will run if the
character is not a digit in the radix as opposed to running only when
the radix is decimal.

There is now a variable *invalid-values* which contains a list of
numbers which are expected to signal an invalid-number error.

Previously a string such as "5 . 5" would be parsed as the number
5.05. I fixed this by adding a check so that an error will be signaled
whenever there is whitespace in the string that is not at the ends.

Close issue #3

cl:parse-integer reads over trailing whitespace, confusing the logic of
parse-integers, where the number of characters read is used to compute
the denominator of the fractional part.

Fixes sharplispers/parse-number#2

Signed-off-by: Utz-Uwe Haus <lisp@uuhaus.de>

The problem is:
CL-USER> (parse-number:parse-number "2.56 ")
=> 2.056

Note that there is a trailing space in the string.

Signed-off-by: Utz-Uwe Haus <lisp@uuhaus.de>

(Fixes one expected failure.)

Covers only the case of .-pos without exp-pos at this point