Fallback to other locations when ComicInfo.xml not at root of archive

[tjarls]

Changed

• Changed: Try using ComicInfo in other location when not present at its canonical location at the root of the comic archive

[tjarls]

What was the difference in terms of memory and performance btw? Just curious what you saw.

Sorry it was late, and did not report on that. The difference is within the statistical noise, so too small to be conclusive. So typically within 1 us, while the whole operation depends on the size of the archive from around 50 us for the zips from the unit tests to 400 with examples from my libraries. My server, where I ran the tests is unfortunately too busy to be precise enough and in some of my tests the order was sometimes even inverted. One example:

Method	Mean	Error	StdDev
TestGetComicInfo2	53.72 us	0.517 us	0.596 us
TestGetComicInfo	53.14 us	0.470 us	0.523 us

[tjarls]

I have updated the benchmarks to compare cases where

1. comicInfo is in the correct location. This is slightly faster than with current version.
2. comicInfo is in duplicated. This would have been the same as 1) but it's slower because the archive is larger
3. comicInfo is not in root but present in subdir. This should be slightly slower. The size of the archive is the main factor in timings and memory usage. The lookup of the right ComicInfo file, regardless of the scenario, is just too small to have any impact.

[tjarls]

Here's the results. 1) is baseline, 2) is "duplicate" and 3) is "outside_root":

Method	Mean	Error	StdDev	Ratio	Rank	Gen0	Gen1	Gen2	Allocated	Alloc Ratio
TestGetComicInfo_baseline	49.72 us	0.437 us	0.503 us	1.00	1	0.4272	0.3662	-	37.2 KB	1.00
TestGetComicInfo_duplicate	52.02 us	0.308 us	0.355 us	1.05	2	0.4272	0.3662	0.0610	38.77 KB	1.04
TestGetComicInfo_outside_root	52.11 us	0.141 us	0.162 us	1.05	2	0.4272	0.3662	0.0610	38.17 KB	1.03

[majora2007]

I have some examples as it also caught me off guard. Let me find them and I'll shoot over via discord.