[rcore] Make FileText*() error handling consistent with Text*() helpers

[Ne0nWinds]

This PR addresses two issues:

1. FileTextFindIndex will dereference a NULL pointer in the following two cases:
   1. the file exists but has empty contents
   2. the file exists but can't be opened (ex. Windows has locked the file)
2. FileTextReplace will completely overwrite the file contents with garbage if:
   1. the file exists but has empty contents
   2. the search string is of zero-length: ""

FileTextFindIndex

If a file exists but has empty contents, LoadFileText returns a null pointer. FileTextFindIndex directly passes this null pointer into strstr, which crashes the program. This is different from TextFindIndex, which correctly handles zero-length strings.

The solution here is to simply check if LoadFileText returns NULL. This makes FileTextFindIndex and TextFindIndex behave consistently with empty strings, in addition to being more robust against failed file loads.

FileTextReplace

If the file is empty or if the search parameter is a zero-length string, FileTextReplace passes NULL directly into SaveFileText. This NULL pointer gets passed into fprintf with the format string %s. This is undefined behavior in C and does different things depending on the compiler and optimization level. MSVC, for example, writes out the literal string (null) to the file (completely overwriting whatever was there before).

Furthermore, in the event that FileTextReplace hits one of these failure cases, it returns 0, so the caller has no way of knowing something went wrong.

FileTextReplace uses TextReplaceAlloc, which already handles all these edge cases gracefully, so the solution is to check if TextReplaceAlloc returns NULL. Doing so makes FileTextReplace have consistent error handling with TextReplaceAlloc.

[raysan5]

@Ne0nWinds thanks for the review!