implement cobra conventions for model identifier conversion

[jotech]

There are different rules for how to deal with special characters like parentheses in the cobra community, compared to what is considered by the sbml importer itself (see opencobra/cobratoolbox#1089).

Currently, cobrar does not convert model ids strings, which could lead to unwanted side effects, for instances, when considering exchange reactions:

Escherichia_coli_str_K_12_substr_MG1655.xml.gz (agora1 model)

mod <- cobrar::readSBMLmod("Escherichia_coli_str_K_12_substr_MG1655.xml.gz") # v1
mod@react_id[grepl("EX_", mod@react_id)][1:5]
[1] "EX_15dap_LPAREN_e_RPAREN_"   "EX_2ddglcn_LPAREN_e_RPAREN_" "EX_4abut_LPAREN_e_RPAREN_"   "EX_Lcyst_LPAREN_e_RPAREN_"   "EX_acac_LPAREN_e_RPAREN_"   

Escherichia_coli_str_K_12_substr_MG1655.xml.gz (agora2 model)

mod <- cobrar::readSBMLmod("Escherichia_coli_str_K_12_substr_MG1655.xml.gz") # v2
mod@react_id[grepl("EX_", mod@react_id)][1:5]
[1] "EX_12ppd_S__40__e__41__" "EX_15dap__40__e__41__"   "EX_2ddglcn__40__e__41__" "EX_3hpppn__40__e__41__"  "EX_4abut__40__e__41__"  

We should implement the conversion of sbml ids according to opencobra/cobratoolbox@d960fbd.
This would improve the support of older models, especially the ones from the agora collection.