renderer: revamp image and shader listing

[illwieckz]

Deep revamp of image and shader listing functions.

• Make it always use Log::CommandInteractionMessage to avoid being truncated by logs.suppression.enabled.
• More std::string, less char* and related C functions.
• More map-based decision (more generic code, and move copy-pasting risks to data instead of logic).
• Added strings for all image formats that I managed to list but were not known before.

A rewrite of the shader listing was already done by me some time ago, the rewrite of the image listing was now done in a similar way but even pushed forward, and the shader listing was then pushed to that same level.

[illwieckz]

As it was already done for shader listing, column sizing is now automatic (no need for hardcoding padding space in endless switch case blocks).

[slipher]

The better way to avoid log suppression would be to use a Cmd::StaticCmd and its Print member function.

[illwieckz]

The better way to avoid log suppression would be to use a Cmd::StaticCmd and its Print member function.

To me the code looks good enough with Log::CommandInteractionMessage to be merged and is already better than current code.

I will postpone or even leave to someone else the task to port those commands to the Cmd::StaticCmd class, this can be done later as a next step.

[slipher]

I will postpone or even leave to someone else the task to port those commands to the Cmd::StaticCmd class, this can be done later as a next step.

Well, using CommandInteractionMessage is something of a step backwards wrt migrating to StaticCmd because for every formatted log you had to pull it out into a Str::Format, whereas with the command you can do a simple replacement of the log function with Print.

[illwieckz]

The whole PR is still a step forward because:

• all the Com_sprintf( buffer, sizeof( buffer ), … are gone,
• there is no hardcoded space padding anymore,
• the padding is now computed from inputs to always be fine,
• the output is not truncated by logs.suppression.enabled,
• some enums are now properly printed,
• etc.

I even consider that migrating to std::string even when using CommandInteractionMessage is still a step forward compared to the char buffer[ MAX_TOKEN_CHARS ] stuff.

[illwieckz]

It now looks ready to me.