Reports can be vague about affected symbols

[bluisblu]

Expected behavior

It would be nice if every specific symbol that contributes to a regressed section in a PR was listed in the report.

Current behavior

Most of the time, symbols (especially functions) are correctly reported. However, I’ve seen some regressions in tssm that only mention sections without specifying which symbols changed. This makes it difficult to tell what exactly regressed especially with the way tssm structures its "wad" files.

Example:

Regressions: 4

⚠️ main/SB/Game/zWad1 - .bss -545 bytes -> 14.81%
⚠️ main/SB/Game/zWad3 - .bss -96 bytes -> 6.56%
⚠️ main/SB/Core/x/xWad3 - .text -316 bytes -> 35.71%
⚠️ main/SB/Core/x/xWad1 - .text -8 bytes -> 0.00%

Progress: 11

📈 main/SB/Game/zWad1 - .text +536 bytes -> 3.80%
📈 main/SB/Game/zTextBox - .text +88 bytes -> 18.14%
📈 main/SB/Game/zWad1 - .text +3 bytes -> 36.08%
📈 main/SB/Core/gc/iWad - .text +604 bytes -> 70.56%
📈 main/SB/Game/zWad1 - .sdata2 +15 bytes -> 6.11%
📈 main/SB/Game/zWad1 - .sbss +54 bytes -> 18.44%
📈 main/SB/Game/zWad3 - .sbss +15 bytes -> 41.60%
📈 main/SB/Game/zWad1 - zCameraReset(xCamera*) +536 bytes -> 99.26%
📈 main/SB/Game/zWad1 - zEnvEventCB(xBase*, xBase*, unsigned int, const float*, xBase*, unsigned int) +4 bytes -> 1.35%
📈 main/SB/Game/zTextBox - ztextbox::deactivate() +24 bytes -> 68.18%
📈 main/SB/Game/zTextBox - ztextbox::activate(bool) +64 bytes -> 84.21%

[SquareMan]

I expect that the issue with this is down to how objdiff handles data diffing. I don't think it usually matches by symbol name but rather by diffing the entire section. I assume this is because a lot of data has anonymous compiler generated symbol names that would never match up. Maybe there's a way though.