Fix Configuration Parser For Size Units

[LucioDonda]

Wazuh version	Component	Install type	Install method	Platform
5.0.0	Configration Parser	Agent	Sources	-

Description

After the implementation #385 it was forced to implement a workaround on the configuration parser in order to correctly handle the size units.

LINK

        // This is a workaround for parsing size units
        if (std::string_view(key) == "batch_size")
        {
            should_parse_size = true;
        }

Initially it was implemented using size_t as the type but that was opening a possible conflict with using size_t as his own (e.g. while parsing threads).
Later another option was creating a custom type but that need to shared the header where it's defined in every module that uses it.
Another option was getting the output type and the parsing type from arguments of the template.

• Design a solution that solves the previous issues.
• Apply changes.
• Correct and / or add any Unit Test needed.

It may also be considered, creating a new public interface to the configuration parser that returns values with a function AsType(), for example:

• .GetConfig("table", "key").AsInt()
• .GetConfig("table", "key").AsString()
• .GetConfig("table", "key").AsTime()
• .GetConfig("table", "key").AsSize()

[jr0me]

Update

• Looked into requirements
• Created branch
• Made ParseSizeUnit and ParseTimeUnit in ConfigurationParser public and static.
• Removed the size and time special cases from ConfigurationParser::GetConfig

The idea is that GetConfig does a simply parsing of a configuration file and casts the values to the specified type. Any additional parsing and conversion must be done separately, a value like 1000MB is a string, and must by retrieved as a string, prior to be parsed as a size in bytes. It still needs to be decided how to improve the ConfigurationParser interface to make these conversions on the go, and to clamp values and set defaults, without pushing too much into the user, but GetConfig will be as simple as possible, maybe additional methods will be provided.