Fix redundant cast-to-int when INI_USE_STACK!=0

[bostjan]

Preamble:
@benhoyt, I understand that this PR is borderline bikeshedding, so feel free to reject it. I mainly wanted to make sure to bring this to your attention, since it might be important to you for other reasons (i.e. other users that are using SonarQube/SonarCloud too may see this annoyance too).

---

Code context (ini.c):

    101: #if INI_USE_STACK
    102:     char line[INI_MAX_LINE];
    103:     int max_line = INI_MAX_LINE;
    104: #else
    105:     char* line;
    106:     size_t max_line = INI_INITIAL_ALLOC;
    107: #endif
    ...
    133: while (reader(line, (int)max_line, stream) != NULL) {

SonarCloud is reporting a "code smell" due to a redundant cast to int at line 133.
This only happens when INI_USE_STACK is true, and not otherwise, as in that case
size_t gets correctly casted to int.

This patch makes the usage of type cast to int depend on the INI_USE_STACK too,
and skips the unnecessary type cast when max_line is already defined as int.

The original SonarCloud report URL:
https://sonarcloud.io/project/issues?pullRequest=221&issues=AYGYdhPZKkhmUWy1nsvP&open=AYGYdhPZKkhmUWy1nsvP&id=snoopy

[benhoyt]

Hi @bostjan, thanks for this. I'm not that enthusiastic about the fix as is, because it adds complication to silence a (false positive-ish) tool warning ... however, what about the following fix -- does that work for your tool?

diff --git a/ini.c b/ini.c
index f8a3ea3..8d3b597 100644
--- a/ini.c
+++ b/ini.c
@@ -100,7 +100,7 @@ int ini_parse_stream(ini_reader reader, void* stream, ini_handler handler,
     /* Uses a fair bit of stack (use heap instead if you need to) */
 #if INI_USE_STACK
     char line[INI_MAX_LINE];
-    int max_line = INI_MAX_LINE;
+    size_t max_line = INI_MAX_LINE;
 #else
     char* line;
     size_t max_line = INI_INITIAL_ALLOC;

[bostjan]

But of course it works like, @benhoyt :)

Clearly your version is more elegant. Not sure why I didn't want to delve into a does that max_size _need_ to be an int with INI_USE_STACK==0? question initially.

Results from a new scan with your patch applied:
https://sonarcloud.io/project/issues?id=snoopy&pullRequest=222&resolved=false&types=CODE_SMELL

I've changed the PR to your suggestion (and added you as a coauthor of the change). Thanks for your time!