Revise trycycler method implementation

[georgiesamaha]

No description provided.

[fredjaya]

Implementation/fixes brainstorm. Ranked according to quality of assemblies output, according to literature and previous testing here. Noting that there is a trade-off between workflow runtime and added complexity vs. quality, Method 4 > Method 6 might be the most suitable and should address #21

$A$ = total number independent assemblies
$n$ = number of subsets per assembler

Rank	Method	Implementation/Notes
1	Trycycler with 12 independent assemblies + plassembler w/ flye	Add two more assemblers and subset 3 reads per assembler
2	Trycycler with $A$ independent assemblies + plassembler w/ flye	Add $A/n$ more assemblers and subset $n$ reads per assembler
3?	bact-builder falls somewhere here, but..	.. has 4 assemblers with 3 subsets each BUT only and always takes the first cluster (ok for their case with no accessory plasmids/chromosomes)
4	Trycycler with 3 independent assemblies + plassembler w/flye	Add a third assembler, no subsetting
5	Trycycler with 2 independent assemblies + plassembler w/flye	Current implementation
6	Use flye assembly if trycyler (2x) fails	Easiest fix at the cost of assembly quality
7?	hybacter	Testing returned less complete assemblies compared to consensus approach

[georgiesamaha]

Tested adjustment of bin/classify_trycycler_clusters.py to filter less stringently (case 3) but that messes up at the reconciliation step due to limited contigs. No go.