Use programs that have been specifically made for bacterial genomes, like Breseq (from FASTQs) or Snippy (from either FASTAs or FASTQs). In the Breseq paper you'll find the reasons why it's wrong to use programs that do not natively support haploid genomes.

In general you could use GATK best practice for both SNPs and indels

If only you care about SNPs

for the SNP calls from Illumina data sets, the pipeline composed of BWA-MEM and Samtools showed the best performance, but Freebayes with any aligner showed an equally high performance for the SNP calls. For the SNPs from Ion Proton data set, Samtools outperformed all others, including TVC. For the indels from Illumina data sets, GATK-HC performed better than the other callers, regardless of the combined read aligner for the pipeline

source Systematic comparison of variant calling pipelines using gold standard personal exome variants

for more and comparison of different tools you can read this paper
Evaluating Variant Calling Tools for Non-Matched Next-Generation Sequencing Data

BBMap's callvariants.sh natively supports haploid genomes; the default ploidy is 1. It's been used extensively tracking SNP evolution in mutant E.coli populations.