I disagree somewhat with your statement that NJ does not provide acceptable phylogenetic information. To me, the main distinction between phenetic (NJ) and cladistic analyses is the scale for which they are appropriate. For questions targeting recent evolutionary questions (i.e. genus and below, depending on the size of the target genus) NJ can be appropriate. You have to keep in mind that phenetic methods are based on various measures of similarity or genetic distance. If homoplasy is not a large concern for your data, then similarity can be a good indicator of evolutionary history, especially for genetic data. However, as you start examining more divergent lineages, the assumption that similarity accurately reflects evolutionary history is not reliable. The other significant limitation is that a lot of information contained in your data is lost with a phenetic analysis when your data is converted to a single measure of similarity (i.e. Fst), which does not happen with a maximum likelihood (ML) or bayesian (BA) phylogenetic analysis. This prevents you from being able to know which character state changes support particular groupings in your tree. But if you have allelic data similarity can be a good indicator of evolutionary history. 

Related your entire genome question, it depends on your data and question. NJ trees are commonly used when people have SNPs because applying a model of evolution and rate of change to SNP data is difficult. NJ is also quick, which can be advantageous when you have a large data set. If your data consists of the entire sequence of a plastid genome, I would use a cladistic approach. Cladistic analyses will also allow you to apply a more complex model of evolution appropriate to your specific data.

Overall, NJ is appropriate for certain questions (recent history) and data types (allelic data), while cladistic methods are preferred for higher level phylogenetic studies and sequence data where you want to extract as much evolutionary information from your data as possible. 

Why do some researchers still interpret results of neighbor joining (NJ) analyses as if NJ were able to provide acceptable phylogenetic information?

(1) Is this because of plane ignorance?

or

(2) Is NJ a valid method for phylogenetic inference if the data are collected across the entire genome of the OUTs?

I would like to gather opinions from colleagues.

I disagree somewhat with your statement that NJ does not provide acceptable phylogenetic information. To me, the main distinction between phenetic (NJ) and cladistic analyses is the scale for which they are appropriate. For questions targeting recent evolutionary questions (i.e. genus and below, depending on the size of the target genus) NJ can be appropriate. You have to keep in mind that phenetic methods are based on various measures of similarity or genetic distance. If homoplasy is not a large concern for your data, then similarity can be a good indicator of evolutionary history, especially for genetic data. However, as you start examining more divergent lineages, the assumption that similarity accurately reflects evolutionary history is not reliable. The other significant limitation is that a lot of information contained in your data is lost with a phenetic analysis when your data is converted to a single measure of similarity (i.e. Fst), which does not happen with a maximum likelihood (ML) or bayesian (BA) phylogenetic analysis. This prevents you from being able to know which character state changes support particular groupings in your tree. But if you have allelic data similarity can be a good indicator of evolutionary history. 

Related your entire genome question, it depends on your data and question. NJ trees are commonly used when people have SNPs because applying a model of evolution and rate of change to SNP data is difficult. NJ is also quick, which can be advantageous when you have a large data set. If your data consists of the entire sequence of a plastid genome, I would use a cladistic approach. Cladistic analyses will also allow you to apply a more complex model of evolution appropriate to your specific data.

Overall, NJ is appropriate for certain questions (recent history) and data types (allelic data), while cladistic methods are preferred for higher level phylogenetic studies and sequence data where you want to extract as much evolutionary information from your data as possible.