In all of your examples, those URIs can (or, well, should) generally be reached by every penetration testing tool. Most tools such as web application security scanners (and often even web application security proxies) implement a crawler (sometimes called a spider).
Crawlers perform link extraction. Link extraction can be benchmarked using a tool such as WIVET. Some tools perform automated form submission, or utilize fault injection. You can sometimes signature penetration testing tools based on this latter information. As an example, Burp Suite submits forms that often contain the name Peter Wiener.
Don't underestimate the power of adversaries to deny or deceive you. Identifying advesary weapons is a common military practice called TECHINT. It could be that an adversary wants you to think that he or she is running a certain tool when they are not -- and it could be that they are noticing that defenses are changing to block their specific tool, so an adversary response would typically be to change things up, i.e., use different tools.
UPDATE (after questioner's comments on why and well as looking at his list again):
Based on the roller-ui URI attack, this is based on an exploit for Apache Roller that relies on OGNL injection (CVE-2013-4212). The exploit was written for Metasploit, but based on some of the other attacks, such as the Solr select check, I can't conclude that Metasploit was used against your website. It looks like it may be a custom tool. Strangely, the JBoss jmx-console/HtmlAdaptor vulnerability appears to be CVE-2007-1157, but that's a strange check for CSRF when compared to the others which are geared towards remote code execution.
SECOND UPDATE -- based on the following two blog posts, I can conclude that Acunetix does check for /solr/select, /server-info and /Account/Register.aspx?ReturnUrl :