Fixes #9386.
Alternative to #9390. Closes #9390.

Motivation / Problem

MacOS said: BOOOO
GCC said: fuck this shit

See #9386 for details and triage. But basically, since we changed a const char * into std::string, more than one compiler went apeshit on us. I present a (possibly temporary) solution: replace it with const std::string_view.

Description

I do not really know why compilers start to track the std::string like this, or let alone why it blows up optimizations (GCC, from what I understand, tracks 50M vars before giving up .. that sounds like an insane alot for the few settings we have). Either way, on a hunch I replaced std::string with const std::string .. no change. So what about const std::string_view? Fixes the issue. Riggghhtttttt....

So to be clear, I have no clue why. I do know it is (most likely) a save operation, as the string itself will be stored in the data-segment, and the string_view will be a thin wrapper around that.

Sadly, this means in a few places where we were replacing const char * with std::string, I revert that situation a bit, as I cheeky use .data() to get this compiling again. By default a std::string_view cannot be casted to a std::string. But I mostly think this indicates places where we should change the std::string to std::string_view instead of the other way around. Either way, I pushed this outside the scope of this PR, to keep this "fix" to its core element.

The commit message:

GCC complaints with:
  note: variable tracking size limit exceeded with -fvar-tracking-assignments, retrying without

clang on MacOS just crashes (with bus error 10, read: out of memory).

This started to happen after introducing std::string to each setting.
As these are static strings that never change, there is no need to
use std::string. Instead, we can use std::string_view for this. This
seems to make both compilers really happy (as they basically still
only store a "const char *" as it was before this commit).

Limitations

• I have 0 clues why this fixes the issue, or why the original code was an issue in the first place. This by no means tells us the root-problem. It just seems to fix the issue so everything is happy again. Any insights to what is going on here exactly would be greatly appreciated.

Checklist for review

Some things are not automated, and forgotten often. This list is a reminder for the reviewers.

• The bug fix is important enough to be backported? (label: 'backport requested')
• This PR affects the save game format? (label 'savegame upgrade')
• This PR affects the GS/AI API? (label 'needs review: Script API')
  • ai_changelog.hpp, gs_changelog.hpp need updating.
  • The compatibility wrappers (compat_*.nut) need updating.
• This PR affects the NewGRF API? (label 'needs review: NewGRF')
  • newgrf_debug_data.h may need updating.
  • PR must be added to API tracker