add lstrip() to lines for parsing

[Leimeroth]

Some package/lammps version change or similar shifts the line
Step Temp PotEng TotEng Pxx Pxy Pxz Pyy Pyz Pzz Volume
which is used as a trigger for parsing by some whitespace, causing problems.
To generally prevent this lstrip() is applied to all lines read from log.lammps

[pmrv]

I'm a bit unhappy with this whole function (not your contribution), because this reads the whole file into memory and iterates multiple times over all the lines. I'm just quickly checking whether I can come up with a single pass solution.

[coveralls]

Pull Request Test Coverage Report for Build 2541304704

• 1 of 1 (100.0%) changed or added relevant line in 1 file are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage increased (+0.002%) to 69.11%

---

Totals
Change from base Build 2536526622:	0.002%
Covered Lines:	11690
Relevant Lines:	16915

---

💛 - Coveralls

[Leimeroth]

It would be possible to simply iterate over the lines directly instead of using readlines by looping:
for l in f:
and only create a df whenever a trigger was found or even read in the whole str up to the end and parse everything using pandas instead of creating 1000s of dataframes.
But I think the logfile does not really matter too much compared to reading the dump file anyway concerning time and memory consumption

[Leimeroth]

Untested draft: #658

[pmrv]

I have definitely produced log.lammps files in the 100's of MB and the old version uses ~1GB of ram for the parse, so I think it's ok to have an eye on it. Thanks for you other PR, I'll benchmark it in a bit.