Try to handle the case where a <c> element is missing an r attribute

[papandreou]

I've run into an Excel spreadsheet where some <c> elements do not have address references. It looks like it's some kind of shorthand where it's implicitly the next cell, and Excel opens the spreadsheet fine. Here's a pretty-printed snippet from the spreadsheet:

  <sheetData>
    <row r="1" spans="1:1" ht="18" customHeight="1">
      <c r="A1" s="1" t="s">
        <v>0</v>
      </c>
    </row>
    <row r="2" spans="1:1" ht="12.75" customHeight="1">
      <c r="A2" s="2" t="s">
        <v>1</v>
      </c>
    </row>
    <row r="3" spans="1:1" ht="12.75" customHeight="1">
      <c r="A3" s="2" t="s">
        <v>2</v>
      </c>
    </row>
    <row r="5" spans="1:12" ht="12.75" customHeight="1">
      <c r="A5" s="4" t="s">
        <v>51</v>
      </c>
      <c s="4" t="s">
        <v>52</v>
      </c>
      <c s="4" t="s">
        <v>53</v>
      </c>
      <c s="4" t="s">
        <v>54</v>
      </c>
      ...

Exceljs breaks with:

(node:3101) TypeError: Cannot read property 'match' of undefined
    at Object.decodeAddress (/Users/andreaslind/work/exceljs/lib/utils/col-cache.js:93:21)
    at value.cells.forEach.cellModel (/Users/andreaslind/work/exceljs/lib/doc/row.js:325:46)
    at Array.forEach (<anonymous>)
    at module.exports.set model [as model] (/Users/andreaslind/work/exceljs/lib/doc/row.js:319:17)
    at model.rows.forEach.rowModel (/Users/andreaslind/work/exceljs/lib/doc/worksheet.js:595:17)
    at Array.forEach (<anonymous>)
    at module.exports._parseRows (/Users/andreaslind/work/exceljs/lib/doc/worksheet.js:592:16)
    at module.exports.set model [as model] (/Users/andreaslind/work/exceljs/lib/doc/worksheet.js:606:10)
    at value.worksheets.forEach.worksheetModel (/Users/andreaslind/work/exceljs/lib/doc/workbook.js:197:23)
    at Array.forEach (<anonymous>)

(The line numbers might be off, as I'm currently running an older version, unfortunately)

I'm not sure the enclosed solution is per spec. Maybe someone else knows more?

And yes, it's actually the same spreadsheet that caused #536 -- I'll follow up if I'm able to find out how it was generated.

[papandreou]

I just learned that the Excel file in question is an export from Xero, which is a piece of accounting software.