Fix the -fstack-protector warning

[ZhksB]

There is a warning appeared when the json library is compiled with -fstack-protector flag using g++ 11th version in Linux Fedora 37.
In our case it turns into error since we use -Werror flag.
The commit aims to fix the issue by adjusting the array size to 8 bytes in lexer.hpp in lines 1553, 1558 and 1563.

Example:
main.cpp

#include <nlohmann/json.hpp>
using json = nlohmann::json;

int main()
{
    json ex1 = json::parse(R"(
    {
        "pi": 3.141,
        "happy": true
    })");
   return 0;
}

Build:
g++ main.cpp -o out -fstack-protector -Wstack-protector -I nlohmann/include

Output:
In file included from nlohmann/include/nlohmann/detail/input/binary_reader.hpp:27, from nlohmann/include/nlohmann/json.hpp:40, from main.cpp:1: nlohmann/include/nlohmann/detail/input/lexer.hpp: In member function 'nlohmann::json_abi_v3_11_3::detail::lexer<BasicJsonType, InputAdapterType>::token_type nlohmann::json_abi_v3_11_3::detail::lexer<BasicJsonType, InputAdapterType>::scan() [with BasicJsonType = nlohmann::json_abi_v3_11_3::basic_json<>; InputAdapterType = nlohmann::json_abi_v3_11_3::detail::iterator_input_adapter<const char*>]': nlohmann/include/nlohmann/detail/input/lexer.hpp:1510:16: warning: stack protector not protecting function: all local arrays are less than 8 bytes long [-Wstack-protector] token_type scan()

The PR fixes the warning.

[coveralls]

coverage: 100.0%. remained the same
when pulling 7daa367 on ZhksB:develop
into 7efe875 on nlohmann:develop.

[github-actions]

🔴 Amalgamation check failed! 🔴

The source code has not been amalgamated. @ZhksB
Please read and follow the Contribution Guidelines.

[nlohmann]

What's wrong with setting --param ssp-buffer-size=4?

[ZhksB]

What's wrong with setting --param ssp-buffer-size=4?

I didn't change it.
I have the error with make amalgamation on Fedora and for this reason I cannot finish the PR. I will try to make an update using a different OS.

[nlohmann]

What's wrong with setting --param ssp-buffer-size=4?

I didn't change it.

I meant: why do you want to change the (correct) sizes in the library code when you can use the parameter above to suppress the warning?

I have the error with make amalgamation on Fedora and for this reason I cannot finish the PR. I will try to make an update using a different OS.

Please see #4275.

[ZhksB]

I meant: why do you want to change the (correct) sizes in the library code when you can use the parameter above to suppress the warning?

I will check. If it works, then there is no sense to change the array size.

[ZhksB]

What's wrong with setting --param ssp-buffer-size=4?

I made the tests and this param works for me. There is no sense in PR more, and I close it.